How a SELECTed Bad Study Became Big News

Introduction
On July 10, 2013, major media headlines and news stories claimed “Too Much Fish Oil Might Boost Prostate Cancer Risk.” Wow, that sure seems fishy given all of the positive health benefits linked to fish oil intake. In examining the study, there are numerous issues that clearly indicate that perhaps the conclusion is wrong, but really a study’s conclusion is only as good as the data used (garbage in = garbage out).
[expand title=”Read More” swaptitle=”Close”]
Data Used was From the SELECT Study
The pedigree of the study source is impressive. It was published in the Journal of the National Cancer Institute and was conducted by researchers from the Fred Hutchinson Cancer Center in Seattle, WA.1 Yet, the data they used is from the much maligned Selenium and Vitamin E Cancer Prevention Trial (SELECT). That is the real problem.

The SELECT study was a very large clinical study that attempted to determine whether vitamin E  alone or in combination with selenium could prevent prostate cancer. Previous studies had shown 50 IU of vitamin E was protective against prostate cancer, but the SELECT study chose to use 400 IU of synthetic vitamin E (dl-alpha-tocopherol) at a dosage of 400 IU. Results showed that the subjects taking vitamin E alone had a 17% higher risk of prostate cancer compared to the control group.

In the new analysis, researchers measured the levels of fats in the blood (plasma phospholipids) and concluded that men with the highest concentrations of EPA, DPA and DHA-three fatty acids derived from fish and fish-oil supplements-had an increased risk of prostate cancer. Specifically, they reported a 71 percent increased risk of high-grade prostate cancer; a 44 percent increase in the risk of low-grade prostate cancer and an overall 43 percent increase in risk for total prostate cancer in a subset of patients with the highest level of these omega-3 fatty acids.

Important considerations of this data are the following:

• This study is not consistent with other studies (discussed below)
• The study did NOT include information or documentation of fish or fish oil intake in the study group. It was NOT set up initially to evaluate these factors, hence its relevance is not as significant as studies designed to specifically determine the impact of omega-3 fatty acids on prostate cancer risk.
• There is no evidence that anybody in this study took fish oil supplements or even ate fish.
• In usual circumstances, plasma levels of EPA and DHA reflect very recent intake and are considered a poor biomarker of long-term omega-3 intake.
• The study did not measure plasma phospholipids on an individual basis and instead pooled data from blood samples collected at enrollment of the study.
• Fish and fish oil ingestion produces a big rise in plasma omega-3 levels in about 4.5 hours and washes out around 48 hours.
• The data may reflect cancer activity rather than a causative association. Without dietary history or documentation of fish oil use there is no way of knowing.

Lastly, the following statement by the authors suggests that they may have significant bias: “There is really no evidence that taking dietary supplements is beneficial to health, and there is increasing evidence that taking high doses is harmful.” Such a statement shows a clear axe to grind in light of a great deal of scientific evidence on the value of dietary supplementation.

A Closer Look at the Reported Results
Let’s take a closer look at the reported results to see if things add up. The bottom line is that they do not. Let’s first take a look at the blood levels of EPA+DHA – the major forms of long-chain omega-3 fatty acids found in fish oil supplements. As Table 1 shows the levels are quite similar among the groups. These blood levels of EPA+DHA are actually quite modest and do not reflect huge levels of fish or fish supplements being consumed. The average EPA+DHA plasma level for men is generally approximately 4%. So, the levels reported here are typical, but a little lower than normal and the ratio of EPA to DHA is also a little lower as well.

Table 1. Distribution of EPA and DHA among SELECT participants by prostate cancer grade (n=2273)*

What the researchers did next was divide cancer patients up by their blood levels of fatty acids and look at the hazard ratio – the relative risk over time – associated with different levels of the various fatty acids (see Table 2). As it relates to EPA, statistical significance was not achieved for total cancer or high-grade cancer (the P value has to be less than 0.05 to be deemed anything more than random chance). For DHA, there was statistical significance. But, again, the levels of DHA are typical of what is found in men consuming modest amounts of fish. However, the level of EPA was lower than that typically found and the ratio of EPA to DHA was also lower. What this may mean is that there may be increased conversion of EPA to DHA in prostate cancer. Though one interesting observation is that the hazard ratio (HR) actually went down in high-grade prostate cancer in the group with the highest level of DHA compared to the next highest group. This suggests that it is not that significant of a factor as one would expect if it was that the higher the level the higher the HR. But, this finding has a P value of 0.09 so no real conclusions can be made as it was probably a random finding.

Table 2. Associations between EPA and DHA among SELECT participants by prostate cancer grade (n=2273)

[/expand]