Obsession as a key to the recent omega-3 fish oil puzzle: omega-3s OK after all?

There has been intense discussion of our report on a recent paper suggesting that there may be heightened risk for prostate cancer diagnosis associated with high levels of consumption of fatty fish (e.g., salmon) and fish oil supplements. The following blog post is written by Jim Waldenfels, who has carried out a detailed analysis of the full text of the original paper by Blasky et al. However, the opinions expressed below are those of the author. They do not necessarily reflect the opinions of The “New” Prostate Cancer InfoLink or its parent organization, Prostate Cancer International.

On July 12, Sitemaster wrote: “Dear Jim: (1) In my experience most men don’t come near to your level of obsession about trying to manage their health.”

With thanks (and apologies) to the creators of the musical My Fair Lady: “I think [he’s] got it! I think [he’s] got it!”. Eliza [Sitemaster] repeats the sentence again and he [Professor Henry Higgins, a.k.a., for this exchange, I] says “By George, [he’s] got it! By George, [he’s] got it!” (From the exchange between Eliza and Professor Higgins in the song “The rain in Spain stays mainly in the plain” with some subtle alterations.)

Not that I haven’t heard the equivalent of that astute assessment of my activities before, chiefly at least once a week at our dinner table.

In all seriousness, having studied this paper, I believe it is more likely than not that the study is picking up not a possible causal connection between increased prostate cancer risks and more marine omega-3 fatty acids coupled with lower omega-6 fatty acids, but rather data associated with increased consumption of marine omega-3s (and perhaps lower consumption of omega-6s) due to heightened concern about developing prostate cancer and awareness of current beliefs about nutrition and prostate cancer risk.  For some of us, including me, concern with lifestyle tactics, including nutrition, to combat the risks of prostate cancer would look at least like strict self-discipline and perhaps like obsession. Sitemaster, I have a hunch that at some level, perhaps even below the level of awareness, you too may be sensing that concern with the risk of prostate cancer, which is perhaps the poorer cousin of obsession (strike that, I mean “high self-discipline” – so much more complimentary), is what is explaining the levels of fatty acids observed in the study. Right up front, let me state that I am far from fully confident of my theory of what is happening, and I do recognize the possibility of some truly increased risk from increased omega-3s/decreased omega-6s as suggested by this paper. I‘m looking forward to some commentary by experts in nutrition and prostate cancer.

What the study reveals to us directly is that men further along the prostate cancer spectrum – a spectrum as reflected in the study spanning: first, some concern but no diagnosis … second, diagnosed prostate cancer that is not “high risk“ … and finally high-risk prostate cancer – consume progressively greater amounts of marine-based omega 3 fatty acids and lower amounts of omega-6 fatty acids. In contrast, the perspective of the authors of the study, and the media take, is that consuming more marine-based omega-3s may increase the risk of getting prostate cancer or its seriousness, and that consuming more omega-6s may decrease these risks, both contrary to conventional wisdom (and most research) until the current and at least two other recent studies. However, at a level a little below the surface details that have been reported, the study also reveals clues to a fair likelihood that this increased consumption is related to the level of personal risk perceived by each participant in the trial and that the increased risks of both getting prostate cancer and developing higher-risk prostate cancer are due to real risk factors behind these perceptions.

Here’s what I think is happening in this trial. (These matters have direct relevance to the details of the trial.) First, the vast majority, perhaps all, of the participants have some heightened concern they will develop prostate cancer, and that supposition is based on the fact that they chose to spend some of their voluntary effort, time, attention, and probably some money in enrollment and participation in a clinical trial for prostate cancer (the SELECT trial), a choice that is quite uncommon for men in the United States. That concern could stem from circumstances as mild as having a friend diagnosed with the disease, or suffering from it, or dying from it, or the concern could stem from more personal and threatening factors.

Second, among all the men with at least that heightened but generalized and non-personal level of concern, many trial participants very likely had even greater concern. We can confidently surmise that having PSA screening results farther from the true norm of less than 1.0 (totally healthy; especially no BPH, no infection, no cancer), or in other words, PSA results closer to exceeding the old biopsy trigger threshold of 4.0 (the upper limit for enrollment in the SELECT trial, coupled with a nonsuspicious DRE), or steadily increasing PSA results, or a post-enrollment suspicious DRE, would increase that concern. We can also confidently surmise that having a “first degree relative” (father or brother) affected by prostate cancer would increase that concern, and the concern would likely also be further heightened if two or more first degree relatives had developed prostate cancer. Having three generations affected by prostate cancer, and having two relatives diagnosed at a young age, especially younger than 55, are additional familial factors that could heighten concern. Simply being African American would likely increase concern in a personal way as the added risk of getting prostate cancer and its seriousness are widely understood in the African American community, I believe.

Third, it’s a safe assumption that men with some concern about prostate cancer risk are more likely to act to try to protect themselves from the risk, and I contend that the likelihood of protective action is typically stronger in proportion to the level of perceived risk. Jonathan’s response of July 15 at 8:26 am to the original discussion is a perfect illustration. His father died from prostate cancer, and his PSA has been elevated for his age. I suspect his father suffered a lingering and often uncomfortable bout with the disease over months or years before dying, which was the case for my own father (and the President’s grandfather, by the way). Unlike the quick death that typifies many cases of heart disease – which, incidentally, is the major risk of mortality for us prostate cancer patients, the prospect of slow decline and death from prostate cancer has an amplified capacity to induce dread and anxiety, and to motivate protective action. Again, Jonathan’s action in seeking out information and taking omega-3s, along with using other health tactics, is an excellent example of the fruits of this motivation, and is so like my own reaction. Here are some examples to show the drastic changes a lot of us make when we confront the risk or actuality of prostate cancer.

As for protection from that risk, back in 1999/2000 when I was diagnosed, about one year before the SELECT trial kicked off, there was a lot of available information stating the possibility, even the strong supposed likelihood, that nutritional tactics could help protect us against prostate cancer or reduce its seriousness. For example, the page of brief recommendations under the acronym CASTLES had been published by the American Cancer Society, as indicated above. Additionally, the well-known financier/prostate cancer survivor and advocate Michael Milken had published a cook book for prostate cancer patients. Internet and prostate cancer newsletters also contained information advocating consumption of marine-based omega-3s. These are just some of numerous resources then available.

By 2001, there was enough enthusiasm to launch the large and expensive SELECT trial for vitamin E and selenium. A search of http://www.pubmed.gov a moment ago for “prostate cancer AND omega 3″ yielded 169 hits, with 141 of them dating prior to randomization for the select trial in July 2001; this was therefore knowledge that could have been available to trial participants. A popular prostate cancer newsletter headlined one issue “Omega 3 Fatty Acids & Saw Palmetto” (separate discussions) in December 2000, and the article referred to ongoing discussions about the benefits of marine-based omega-3s, risk of plant omega-3 (ALA or α-linolenic acid), risk of fish contaminants, and safe access to marine omega-3s. My impression is that marine omega-3s have continued to grow in popularity since that time.

Is it not likely that many of the motivated participants in the SELECT trial would have been taking marine omega-3 fatty acids? Is it not likely that they would have been more likely to take them and to take higher doses (including fish) as their risk factors increased? I’m convinced the answer is yes, but that needs to be tested. If so, then I suspect the associations reported in the study between higher levels of marine omega-3s, etc., are just caused by higher consumption without causing higher risk. In other words, if that suspicion is true, we are seeing a spurious association and not a causal relationship.

Here are key details of the current study. Two groups were set up, all participants in the SELECT trial, comprising men who, as required by the trial’s eligibility rules, did not have prostate cancer at the start of the SELECT trial, as indicated by their histories, by PSAs not exceeding 4.0, and by the absence of suspicious DREs. One group, 834 in number, was basically all SELECT trial participants who were diagnosed with prostate cancer through July 31, 2009, about a year after the unblinding of the SELECT trial. The “control” group, 1,393 in number, was a randomized selection of men who were not diagnosed with prostate cancer through July 31, 2009, matched by race (black, others) and age, with fairly close equivalence happening to be achieved for education (somewhat higher among those diagnosed), smoking, body mass index, alcohol consumption, finasteride use, aspirin use, SELECT intervention assignment, and history of diabetes.

However, there were two characteristics for which the groups were not well matched at all: the first was PSA level, and the second was proportions in the groups with first degree relatives diagnosed with prostate cancer. That is the smoking gun (at least in my opinion)!

Regarding PSA, we see this:

The results also show that nearly twice as many men in the group that eventually was diagnosed with cancer had at least one first degree relative with prostate cancer, with more than three times as many having two or more such relatives! Again, that is a sharp tilt!

In short, in view of these two key risk factors, we can see that risk was markedly higher in the group that eventually developed prostate cancer. Moreover, both factors were clearly evident to each man, and both were well known and publicized at the time as risk factors. I contend it is a safe bet that the vast majority of men who were at higher risk knew it. I contend it is a safe bet that many of them were motivated by that higher risk to consume more omega-3s. If those contentions are sound, it is quite likely that the results reported here of higher risk associated with omega-3s are simply due to consumption patterns and not because omega-3s are increasing risk! I hope that’s true, as it would leave me with an abundance of research evidence suggesting that marine-based omega-3 consumption decreases prostate cancer risk.

Finally, let’s consider the differences in the amounts of marine-based omega-3s consumed that are being associated with the greater likelihood reported in this study of both diagnosis and “higher risk” disease:

These differences strike me as small. The study tells us that these differences are significant, but that means “statistical significance.” It does not prove medical significance. Moreover, even small differences are more likely to be “statistically significant” when the number of study participants is large, as is the case here. Is it not likely that the substantial differences in the PSA and first degree relative factors, displayed above, are associated with substantially more perceptions of higher risk in the group that was later diagnosed, and that those substantially greater perceived differences in risk therefore account for the relatively small differences in consumption of omega 3s? I suggest that the same line of thought applies to the “higher-grade” cancer results.

Because the data are all available and no doubt computerized, it would be worthwhile and probably technically easy to match marine-based omega-3 results for the cancer group above with results for a group of non-cancerous SELECT trial participants, but this time matching for both PSA level and first degree relatives. In other words, let’s do the matching that was not done in this study. That would presumably equalize average perception of prostate cancer risk in the two groups. My hypothesis is that such a study would result in no significantly increased omega-3 consumption by the diagnosed group. If significant differences still remained, which I doubt, with higher consumption by the diagnosed group, would that not suggest that something other than perception was contributing to the greater consumption and associated risk, such as the causal connection suggested by this study? Again, I doubt that is the case, but let’s put it to the test.

In short, the findings of the study appear to be an artifact and, if so, should not bear on our consumption of marine-based omega-3 fatty acids, but it should not be difficult technically to put this assessment to the test.

18 Responses

  1. Jim:

    Thanks for taking the time. What is the measurement used in the chart setting forth “amounts of marine-based omega-3s consumed”? Are those numbers a percent of total fatty acids? Does the paper state actual serum EPA and DHA levels in the respective groups? Is the paper suggesting that high levels of EPA/DHA are potentially problematic, or only that high levels of EPA/DHA relative to total fatty acids are potentially problematic? (By the way, my uneducated reaction to your main thesis — that the “worried guys may eat more fish/fish oil” is: Could be, but who knows?)

  2. Jim,

    The difference in the omega-3s between the groups sounds insignificant.

    Any idea on what a normal blood level would be for the average American male and one who takes, say, a tablespoon of fish oil per day?


  3. Thanks for the thorough analysis Jim. Sounds spot on to me. And should be testable as you say. Maybe you should send it to the guys who did the study?

    Here is another thought. Guys taking omega-3 are probably more likely to take have been taking other supplements as well. Maybe one of those is really dangerous?


  4. Jonathan:

    Re: Your question about the amounts of marine-based omega-3s consumed, etc.

    Those numbers are percentages of total fatty acids. Thanks for asking; I should have stated that. Under “Data Collection and Laboratory Methods,” the paper states “Fatty acid composition is expressed as the weight percentage of total phospholipid fatty acids.” This is where I wish I had taken advanced biology; heck, I’d even settle for better recollection of the biology courses I did take. Apparently there are many other phospholipid fatty acids; perhaps one of us can explain the main ones common in humans beyond the ones mentioned in the paper. In addition to the marine omega-3s in the table above, for the “no cancer” group, ALA contributes, according to the paper, 0.13, linoleic acid contributes 19.03, arachidonic acid contributes 11.40, trans-fatty acid (TFA) 18.1 contributes 1.41, TFA 18.2 contributes 0.20, and TFA 16.1 contributes 0.21.

    “Table 2” in the paper also provides the “95% confidence interval” for each mean percentage value. For those not familiar with statistics, that interval is the estimate of the bottom and top of the range that would include the true value 95% of the time; the mean value is “true” for the study but may not be the true value if an extremely large number of people participated in the study. The intervals in this study look fairly tight to me for most values.

    You asked: “Does the paper state actual serum EPA and DHA levels in the respective groups?” No, and, as a patient, that frustrated me too, both at first reading and now. That is a very interesting piece of information for patients, but maybe not so much for researchers. I would like to know. I tried earlier to see if I could derive it from other numbers, but I don’t see that is possible based on what is in this paper.

    You asked: “Is the paper suggesting that high levels of EPA/DHA are potentially problematic, or only that high levels of EPA/DHA relative to total fatty acids are potentially problematic?” I really appreciate your asking these questions that help us all understand what is going on. I had not appreciated that distinction until your question. The authors mention “high concentrations” and “high intakes” at various points. Personally, I’m thinking that a high percentage of the total probably echoes high consumption closely, but it would be nice to know, and I really do not know, though the paper suggests that. Here’s a pertinent sentence from the Discussion section: “Expressing fatty acids as weight proportions … could create spurious results because an increase in the percentage of one type of fatty acid requires a decrease in others …; however, given the very low concentrations of omega-3 PUFA, it is unlikely that their variability, which is strongly related to dietary intake, would be strongly affected by proportions of other phospholipid fatty acids.” It also notes, “Factors other than diet affect proportions of phospholipid essential fatty acids. For example, in feeding studies a low-fat diet modestly increases the blood proportion of long-chain omega 3 PUFA …; however, these effects are small compared with those due to supplementation ….” Some of the studies listed as specific reference sources may be helpful to us.

  5. I wouldn’t have expected anything less or any more sound from friend Jim. I’ll side with his conclusions every time. As we say in the military, “Our Country, Right or Wrong,” so say I about Jim.

  6. Jim

    You mentioned your acquired taste for soy. Depending on which school of thought you hold regarding the impact of sex hormones and prostate cancer, one consideration may be soy’s estrogenic impact. Some studies suggest that low testosterone is more common in men with prostate cancer and estrogen naturally rises in older men. So less soy may be justifiable along with maca root (an estrogen blocker) and fenugreek tea (a testosterone booster).

  7. Jim –

    Regarding the following sentence you quote:

    “Expressing fatty acids as weight proportions … could create spurious results because an increase in the percentage of one type of fatty acid requires a decrease in others …; however, given the very low concentrations of omega-3 PUFA, it is unlikely that their variability, which is strongly related to dietary intake, would be strongly affected by proportions of other phospholipid fatty acids.”

    I am just wondering about this: One guy eats a generally healthy, largely vegetarian diet, but eats fish a few times a week and supplements. Another guy eats the same amount of fish and fish oil supplements, but also eats a bunch of bacon cheeseburgers and fries. Maybe they would have the same absolute EPA/DHA levels, but the cheeseburger guy would have a bunch of other fatty acids, at higher levels, so for him the percentage of EPA/DHA would be lower. And for purposes of this study, that guy would be in the “lower risk” group? Maybe I should stop at McDonald’s on my way home today?

  8. I also wonder how constant EPA/DHA levels are. I personally plan to test this (sample size of one) by ceasing taking any fish oil pills for 30 days and then doing an essential fatty acid profile test.

  9. Incredible job Jim! I had come to the same conclusion you have.

    Again, great analysis.


  10. Terry,

    Do you have any studies on maca being an estrogen blocker?


  11. My computer crashed within hours of sending the article. In a way that’s fortunate – I launched it, but I’m now going through the time consuming process of selecting a new one. I’m working at our nearby library.

  12. Hi Richard,

    I’m responding to your post of July 16 at 12:14 pm.

    It’s certainly plausible, I think likely, that participants were taking other supplements too, including some that were probably harmful. One of those on my list of suspects would be ALA, alpha-linolenic acid, which is plentiful in flaxseed oil and some other sources. The data in the study, however, showed no elevation in ALA levels.

    One way or another, I hope the authors and media will be alerted to what looks like a major flaw in the study. I may help with that.

    Take care,


  13. Hi to all who are participating on this thread and thanks.

    Special thanks for your kind words to Chuck, Jonathan, Richard, and Stan.


  14. Doug,

    Re your July 16 12:06 PM question, I don’t know what the normal level is and was unable to find out after brief searching. I would like to know too.


    Hi Terry, I’m responding to your July 17 post at 1:30 am. That connection of soy and an added risk due to low testosterone may hold up for some men some day, but I’m solidly in the other school of thought. I surely hope I’m right, because I am constantly wearing a Vivelle estradiol patch to protect bone mineral density while on Lupron. Of course DES — oral estrogen — to combat prostate cancer was the basis for the Nobel Prize, as I recall it.

  16. By now many of us have probably seen some detailed responses from critics of this study, including Dr. Anthony D’Amico, Dr. Mark Moyad, and Dr. Charles “Snufffy” Myers. (Dr. Myers published a video blog on the subject on July 24.)

    They emphasize that this study finds data that associates higher-risk prostate cancer with omega-3 fatty acid intake, but with no evidence at all of a causative relationship. Dr. Moyad concludes that he has to stop writing so he and his wife can go out and have some sushi! (My kind of guy!) Among many points made by Dr. Myers is the key point that a far superior population study, like this non-randomized population study, found a 24% reduction of metastatic prostate cancer risk for each half-gram of fish oil intake (fish or supplements)! The study was part of the ongoing review of, then in the early 2000s, 47,822 health professionals as part of the Health Professionals Follow-Up study run by the Harvard School of Public Health. Dr. Myers noted that this one study runs sharply counter to a monumental number of studies of fish oil and prostate cancer that are favorable to fish oil consumption, provided, of course, that the fish/fish oil is not contaminated.

  17. Such is the fickleness of Science. There was a time when eggs were just so bad because it will raise your cholesterol. Today, it isn’t, then it was, then it wasn’t.
    And Theory of Evolution believers want to make a field such as Science their god? A field that thrives on errors and fickleness?

  18. Ummm … Excuse me … On behalf of the small population of the human race that happens to think that the theory of evolution explains a vast amount of information about the way living things have gone about their lives over the past several million years … few of us have ever suggested that Science is a replacement for God.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: