Thursday, August 08, 2013

Omega-3 / Prostate Cancer

Hello! We all know that 'tomorrow' means 'four days' 'more than a week', right?  Sorry about that, I've been sick with some persistent headache thing. It's most likely the zombie virus from WWZ. Nothing serious.

I am assuming that all readers have heard that a new study shows Omega-3 supplements cause prostate cancer (OMIGOSH).  If not, you must be a healthy individual that ignores media hype; I admire you.  You can catch up on this link from the CNBC and this link from the Telegraph as examples of the media nuttiness.

In the supplement industry, this scandalous news has caused no end of drama: cancelled orders, panicked emails, phone calls from customers and urgent requests for a 'position statement' from industry partners.

Not everyone is wigging out, but even a calm person could read '71% increased risk of aggressive prostate cancer' and start giving their Omega-3 capsules the side-eye. Considering the well-documented benefits of Omega-3 intake, which include cancer reduction, and the fact that prostate cancer is a vicious killer, it's not a small deal.

At uni, we learn to 'critique' such research. This sounds sophisticated, but it's not really. It's more like, if this latest study is the couch, can we eviscerate it? 

In this case there are multiple sofas: the 2012 study which suggested that higher serum Omega 3 is correlated with increased chance of prostate cancer, the previous study which had suggested the same thing in a slightly different way, the media commentary and the many critiques floating around online. The interwebs is fluffy as hell right now.

I'm going to have a go at the two main cushions: study 1 and study 2.

Study 1: 

The first study (2010) was a subgroup analysis from the Prostate Cancer Prevention Trial. This trial had the advantage of being both large (18,882 men) and long (7 years).  Importantly, this was a placebo-controlled drug trial for Finasteride - a pharmaceutical which reduces benign prostate enlargement. Half the group were taking it, and half were on placebo. So, right up front we have a massive confounding variable. Throw a drug into the mix and things get complicated.

At study completion, all men that had not already been diagnosed with prostate cancer were given a prostate biopsy... (erk... I have a friend that gets one every year because he has low-grade cancer and it sounds traumatising.. I can't read 'prostate biopsy without having a moment of painful sympathy in my soul..).. even if they were symptom-free, and then all with evidence of cancer (1658 individuals) were matched for age, treatment and family history of cancer with a group that didn't develop cancer.  This is a relatively robust method for determining what might be different between those groups.

The analysis then examined group-differences in inflammatory-related serum phospholipds including omega-3's, omega-6's and trans-fats.

To everyone's gobsmacked surprise it was found that those with the highest grade prostate cancer had the highest plasma levels of the Omega-3 DHA and that those with higher levels of trans-fats in their blood were less at risk of high-grade prostate cancer.  These results were exactly the opposite of what was expected based on many previous studies.  I imagine, in the stats room, there was a moment of silence, and then someone in the lab took the call, dropped a petri dish and whispered 'Oh My God... what have we discovered???' And then they went to the pub.

For those that like stats, the quartile 1 vs. quartile 4 (of serum-DHA) odds ratio was 2.50. That is, between the lowest 25% and highest 25% DHA levels, men had 2.5 times the risk of developing high-grade prostate cancer. 

Other findings from the study were:

- No association with any fatty acid and low-grade prostate cancer. Iiiinteresting. I'll talk about this in   a minute.

- The association with high-grade cancer was only observed for DHA and DHA+EPA, but not EPA on its own.

- Out of the 125 with high grade prostate cancer, 80 (64%) were from the finasteride group. This contrasted with the low grade cancer group, where 60% came from the placebo group. The authors gloss over this inconvenient correlation by, well, noting it:  'high grade cases were... more likely to have been randomized to the finasteride arm', then ignoring it for the rest of the discussion.

Study 2:

This study is the one the media are going nuts over. It is lead by the same researcher (Dr Brasky) and is a nutritional intervention study, the primary purpose of which was to examine the effect of selenium and/or vitamin E supplementation on prostate cancer risk. Again, this was a large and robust study (35,533 men, 7-8 years).

Note: Omega-3 was not one of the nutrients manipulated.

Much like the previous study, a subgroup with cancer (834) were matched with a control group without cancer and the analyses were performed. In this case, the control group was larger (1364). In contrast to Study 1, it does not seem that everyone submitted a prostate biopsy... (squirm).  Therefore it is reasonable to assume that this study may not have identified all cases.

What they found are the scary results that have been reported ad infinitum in the media:
44% increased risk of low grade prostate cancer
71% increased risk of high grade prostate cancer
43% increased risk of any type of prostate cancer
between those with the lowest 25% serum Omega-3 (all types) and those with the highest 25% (Quartile 1 vs. Quartile 4).

Buuuut. Although honestly reported, I see no immediate need to go stick plastic horns on your omega-3 bottle. Let's break. it. down.

Firstly, it's important to really think about what the numbers mean. What we are looking at is the cancer risk in a subgroup where 61% of the group is known to have prostate cancer. This does not represent the usual free-living population.

I had some difficulty figuring out exactly what the percentage of prostate cancer incidence is in New Zealand males. An extensive search of the uni databases yielded some cryptic figures relating to increases over time and so on. On consultation with my classmates, we concluded that based on MoH and cancer society stats such as this and this, it is somewhere below 1% (of male population) new cases reported per year, but we don't know what a cross-section would look like at any one time. This figure seems low, so if anyone does know, please pipe up. Make me wrong, I'm ok with that, plus I have the zombie virus and don't care.

In any case, the real level of prostate cancer is a lot less than the subcohort and it's crucial that people realise that the results do not represent increased risk over average risk. 

In order to get around the fact that their group was already abnormally cancerous, the researchers have used a hazard ratio. That is, they took the incidence of cancer in the total subcohort (those with and without cancer) for the bottom 25% of serum Omega-3 and called that '1'.  Then they compared that to the incidence in the top 25% of serum Omega-3.  I spent quite some time enhancing my forehead wrinkles by trying to re-do their analysis before figuring out that it's impossible because they do not report a number for how many were in the bottom 25% serum Omega-3 in the subcohort without cancer. 

Now, I'm no statistician (since Lasik, I don't even wear glasses), but it is well known that the hazard ratio is not considered very reliable, even in the context of a controlled clinical trial, which this was not. As I don't have the nous to argue the point eloquently, let's just leave it at that and address the other points that need consideration.

1) The Omega-3's were measured as a percentage of total lipids, which the authors acknowledge is a limitation. That is, we are not looking at absolute levels.  So, if a person has high plasma lipids generally, they could feasibly end up in the bottom 25% of serum Omega-3 even if their measured level of Omega-3's is higher than someone in the top 25% (make sense?).  

2) You may have heard this a lot >>>>>>  correlation does not equal causation. Say there is a correlation between high serum Omega-3 and prostate cancer. We really need to know why those people have more Omega-3 in their blood and to my eternal annoyance, the researchers did not look into this and that makes me want to punch them something.

in the face.
(You know, I think this zombie virus is making me grumpy.  Violence is wrong. Always wrong. Don't hit, just hug.).

Would it have been too much to ask those with high-grade prostate cancer if they were taking Omega-3 supplements?

Neither do we know if they ate more fish than the low serum Omega 3 group. We have no idea. There's a big hole in the middle of their theory that Omega 3 intake is related to prostate cancer, and that is, intake was not measured or even estimated.

Here are my thoughts on reasons why high plasma Omega 3 and prostate cancer might co-occur:

- Prostate cancer is more common in the obese.  An obese person may be trying to lose weight. Weight loss diets generally contain more fish, and recommend fish oil supplements. Eating more fish may also increase the burden of mercury and other heavy metals. Is this related to prostate cancer?  I don't know.

- A person with prostate cancer may start to feel unwell, or worry about their health. They consult Dr. Google who, besides being cheaper and more punctual than their regular GP, is likely to recommend eating more fish and taking fish oil supplements. So they eat more fish because they have cancer, not the reverse.

- The disease process may itself affect biomarkers. This is not unheard of. For example, a viral infection can artificially elevate iron levels.  Cancer quite often causes weight-loss, which liberates fatty acids into the blood stream.  Is it not conceivable that this could mess with the lipid profile?

- The Confidence Intervals (CI) reported in the study were huge. How does that work?  Say we take the most dramatic statistic (71% increased chance of high-grade prostate cancer). The 95% CI was 1.00 to 2.94. This means that the researchers can say with 95% certainty that the actual increase in risk is somewhere between 1.00 (no increase) and 2.94 (194% increase).  Any result that touches '1' (the 'line of no effect') should always be a bit suspect.

- Polyunsaturated fats are quite fragile and prone to oxidation within the body.  There is not a lot of research on this, but there is the possibility that if you increase levels of PUFA's, with inadequate antioxidant intake/levels, you might increase the levels of damaging free-radicals.  So, a bad diet (say, low in veges) + Omega 3 might actually stimulate production of free-radicals.  I think this is an area that will attract more research in future.

- Differing individual responses to Omega 3's, and particularly DHA, are showing up in all sorts of studies which have investigated genotype.  The different ApoE genotypes, of which there are three major types (E2, E3 and E4) differ in the way that fats are metabolised and this affects response to typical applications of Omega 3, for example to assist high cholesterol or in dementia prevention. It would be very interesting to know if the high-grade cancer group were of a particular ApoE genotype. 

I also had a random thought about why the study 1 found a correlation with high-grade cancer but not low-grade. This is entirely out of my brain, so try not to judge. Just throwing it out there.

Substances that affect gene activity through a process called methylation, 'may' enhance a disease process that is already underway. I first heard of this when investigating mandatory folic acid fortification of food. Although there are many 'pros' to fortification, including cancer prevention, one of the 'cons' is that it can cause faster progression of cancer, if there is already cancer present.  It seems to support whichever process is already happening - whether that process is healthy or diseased. 

It could be that if prostate cancer is already present, but unknown, then Omega 3 encourages it into a more aggressive form. Like I said, that's just a personal theory.

Lastly, neither of these trials were designed to test any hypotheses related to Omega 3. The Omega 3 analyses were piggybacked onto trials designed for other things, and that adds a layer of uncertainty to the results.

What do you think?  Has this caused you any problems with your clients or family?  It's tricky when a new study throws a spanner in the works, and not helpful when the mainstream hacks go to town on it.  Hopefully this post has shed a little perspective. If you have any questions, just comment me and I'll do my best.



  1. Hey, you are alive! This is good stuff, your brain is working but what is up with the headaches? Imma gonna make your buddy Kek tell you off about not getting this checkeed out.

    1. She already did! She Facebook spanked me. Thanks for your concern, I'm fine and thankful to not even have a prostate. X

    2. Oh dear. Am I "that" friend? :D

  2. Rosslyn6:46 pm

    i love it intensly when you get sciency. It saves me the effort. I work in a health food shop and there are lots of customers asking about this so its good to know where to send them. Denise Minger also wrote on it on Marks Daily Apple

    1. Thanks! That's the first study ('Study 1') she is talking about though. I think that was written before Study 2.

  3. Nice work Sara. It definitely helps to have someone actually think deeply about this so that I don't have to. My current lifestyle doesn't lend itself to thinking deeply very often!

    1. I am surprised you can think at all. Every parent has my full respect. ;)

  4. I haven't heard of this study (maybe it hasn't made the rounds in the US news yet? Or maybe I get the majority of my news via facebook and twitter...), nor is this a topic I'm knowledgeable in, but your analysis was still really fascinating to read.

    1. There was just a 1 minute segment on the news here (NZ), so that didn't cause a LOT of panic. The main concern has come from the bodybuilders, the paleo crowd, that take heaps of Omega-3 (often 20g or more a day) and those that are ill - these people spend time on forums and talk to each other. I'm sure these researchers will do another study to prove their point and it will flare up again, but I'd expect the online backlash has calmed most people down.

  5. does a higher level of omega 3's in the blood even indicate that test subjects consume higher amounts of O-3's? Maybe it just means that their body isn't absorbing it as well as some people's do. Or maybe, like cholesterol, the higher blood level isn't based on consumption but on having more damage to the body that cholesterol is being transported to in order to fix.

    1. That is a good point. Omega-3's do rise in a dose-response relationship but this does not mean that intake is the ONLY thing that might affect levels. The body is complicated. Like you say, it can adjust absorption and transport as needed.


I love to hear from you! Tell me what's in your brain, your heart or your dinner plate :D.