Mercury in Fish NOT Harmless, Regardless of Cysteine, Selenium, EPA or DHA! Plus: No Cardioprotective Effect of Omega-3 in Men With Higher Hair Mercury Levels

Image 1: Nice! Luckily nothing you will catch everyday, because if you ate this little bastard, a Tile Fish from the Gulf of Mexico, everyday, you could - in the worst case - be consuming 933µg of mercury with every 250g serving!
"Mercury from fish is not a problem, because you get plenty of selenium to counter it... moreover it's mostly protein bound, already..." - Another Myth Busted!? I must admit, I did believe (without ever checking scientific references) the common mantra that the mercury (Hg) content of fish would not actually be a problem, as long as there is enough selenium (Se) in the fish to "buffer" the Hg load. Now, this certainly makes sense and even very recent studies confirm that the effective uptake is reduced with higher Se:Hg ratios (e.g. Calatayud. 2012). Moreover, the notion that selenium exerts a protective effect is bolstered by data from various indigenous populations in the Brazilian Amazon (Lemire. 2011).

Cysteine, Omega-3 & Selenium? Won't help!

Unfortunately, a recent study by a group of scientists from the Arcachon Marine Station in Acachon, France, does now remind me why I have made it a rule over the year to question every conventional wisdom regardless how logical it may seem (Bourdineaud. 2012). The researchers fed a group of mice diets that contained either 4.88% fishmeal powder that had been produced from the flesh of H. aimara fish that had been caught in the Sinnamary River in French Guiana and contained 5µg Hg/g or a control diet which had slightly less protein (14.2% vs. 18.1%) and contained higher concentrations of EPA(10x), DHA(>30x) and DPA (>5x) - obviously right from the fish.
Figure 1: Fatty acid composition of the diets (left) and breakdown of the omega-3 part of the diets (rel. to total PUFA content - right; data calculated based on Bourdineaud. 2012)
In addition, the fish diet contained methylmercury in its purportedly less toxic largely peptide bound form, methylmercury-cysteine (MeHg-cysteine), while the mercury the scientists had added to the control diet was the purportedly more toxic salt form of mercury, i.e. methylmercury-chloride (MeHgCl).
Which fish contains how much mercury? I knew you would ask this and in essence it is impossible to answer without analyzing the very same fish, because as we are about to see, even the same species from the same fishing ground won't do.

Figure 2:  Mean (bottom axis!) and max (top axis!) mercury content (mg/kg) in fish (based on FDA Monitoring Program. 1990-2010)
Now, I would be a hilarious smartass if I left you with this "you never know" statement, but would still advice you to regard the following information as very broad estimations and heavily generalized categorizations:
  • the worst offenders: Mackerel, King Shark, Swordfish & Tilefish (from the Gulf of Mexico) with mercury levels in the 1,000µg/kg range - 250g of those and you are on par with the mice in the study
  • examples from the rest of the pack (see figure 2): It is plain to see that even fish with a relatively low mean mercury concentration such as Pollock (mean: 31µg/kg) can be laden with mercury, if you just pick the wrong one (max: 780µg/kg!)
Regardless of in some cases 20x higher outliers, you are probably on the safer side of things, when you pick one of the fish / shellfish that are on top of figure 2 and thus have the lowest mean mercury concentration.

How much did the mice consume? With  253 and. 237µg/kg in the MeHgCl and fish diets the mice in the study at hand consumed ~1µgof mercury per day this corresponds to a human equivalent dose of approximately 3.2µg/kg or 263µg/day for a 80kg adult.
Next to the aformentioned selenium argument (the selenium content of the fish diet was likewise higher 480 vs. 300µg/kg), the presence of MeHg-cysteine instead of MeHgCl and the healthy fish oils, are arguments #2 and #3 in the unquestionably convincing "mercury from fish is not a problem" argument.

It takes 8 weeks of mercury expose for the mice to go havoc - only from fish, though!

The mice were maintained on the diets for either 29 or 58 days. At the end of the exposure period, mice were subjected to an open-field maze test, in order to quantify anxiety levels, and to a Y-shaped maze test, to assess cognitive ability. Thereafter, the rodents were anesthetized and tissue samples were taken. Here are the main findings:
  • within the first 10 days of the feeding period, the mice on the Hg containing diets gained  weight faster than rodents on a non-Hg control diet - 4%  and 7.4% more weight gain in the MeHgCl and Fish group, respectively; afterwards the weight development was identical
  • both Hg diets lead to significant increases in serum and tissue MeHg with the kidneys being the "preferred" storage place with a tissue concentration of 7.3 and 6.8 mg Hg/g in mice fed the MeHgCl and fish diets, respectively (17x and 16x higher than in controls); there was a statistically significant inter-group difference only in the striatum, which accumulated ~30% less methylmercury in the fish group compared to the MeHgCl group
  • significant behavioral abnomalies did only occur on the 2nd test at the end of the study period (day 58) and were exclusive to the Fish group, which also exhibited an increased dopamine metabolic turnover in the hippocampus
In the end, there is little to add to the scientists somewhat disillusioned conclusion that despite the fact that they had had good reason to assume (like you and I ;-) that the mercury induced metabolic and neurocrine perturbations in the Fish group "should appear less severe than that observed with the MeHg-containing diet [..] the present study" falsified the original hypothesis and suggests that rather than being less toxic, the peptide bound MeHgCysteine in fish is even more toxic than its chloride bound counterpart.

"Mice are nice, but what about men? I am sure know fish oil protects us!" Not really, no...

Another of the pieces that's still missing to get at least a preliminary grasp of the fish oil, selenium, mercury-toxicity puzzle, comes from a recent study that's been conducted at the University of Eastern Finland in Kuopio, and in the course of which the scientists analyzed the relation of mercury exposure (as quantified by hair mercury levels), long-chain poly-unsaturated fatty acids (LC-PUFA = omega-3) levels and individual risk of CVD, in general, and sudden cardiac death, in particular, in a group of 42-60 year-old men who had been free of any adverse cardiovascular events at baseline in 1984-1989 (Virtanen. 2012); and the results Virtanen et al. present in a paper in the July edition of the free medical Journal PloS One are astonishing, to say the least:
  • of the three long-chain polyunsaturated fatty acids, EPA, DHA and DPA (=docosapentaenoic acid), only the latter, i.e. DPA, correlated significantly with the absence of sudden cardiac death within the time to the follow up (p < 0.01)
  • the by far best predictor of whether or not the study participants would pass away before their time was yet the hair mercury content, which was 53% higher in those unlucky 91 patients who died from sudden cardiac death, than in the "survivor" group (2.85µg/g vs. 1.86µg/g)
Before we take a closer look at how this translates into the calculated hazard risks, I do yet feel inclined to draw your attention to some more basic, and not statistically processed baseline characteristics of the participants with the highest (4.96–15.59%) serum LC-PUFA values.

Don't deduce from pairs of associations!

A brief lesson in interpretation of scientific data - If A & B, and A & C, then B & C... NO!

Actually this thing about associations and logical reasoning is nothing extraordinary, but I thought it may be worth reminding you not to make the false assumption that  "if A is associated with B and A is associated with C, then B must be associated with C, as well", or to give you a more concrete example: If people with high LC-PUFA levels have higher incomes and people with high LC-PUFA levels have higher mercury levels, then people with higher mercury levels should also have higher incomes"

I see, now you are laughing, but I bet, everyone of us has once fallen for a similar mistake, esp. if the result of this falsely applied deduction was in support of your original hypothesis.
The study participants with the highest long-chain omega-3 levels in their blood also had the highest...
  • physical activity (borderline significant p = 0.06)
  • income (p < 0.001) and eduction (p = 0.01)
  • fish, fruit, berry and vegetable intakes (p < 0.001)
  • the highest hair mercury concentration (p < 0.001)
  • the highest alcohol intake (p < 0.001, and 53% more than those w/ 1.7-3.9% LCPUFA)
  • the highest rates of coronary heart disease in the family (p = 0.03, but only 6% difference total)
Despite the fact that higher mercury levels in the had were thus obviously associated with higher omega-3 levels in the blood, it would be preliminary to assume that all other of these variables, such as a higher income, or the physical activity would also be associated with higher mercury levels. And in fact, the exact opposite is the case,...
  • higher income,
  • higher education,
  • higher fruit and vegetable intake and
  • higher physical activity
... all of which were also associated with higher omega-3 levels in the blood were statistically significantly associated with lower mercury levels!

Mercury, fish oil and heart disease a marvelous triumvirate 

Let's get back to the harzard ratios and how fish oil intake and methylmercury intoxication interact in terms of the sudden cardiac death risk of the middle-aged (mean age at baseline 52.1 years) study participants.
Figure 3: Hazard ratios relative to lowest - adjusted for age and examination year (model 1),  adjusted for model 1 and body mass index, pack-years of smoking and alcohol intake (model 2),  adjusted for model 2 and hair mercury content (model 3); and hazard ratios associated with each 0.5%  unit increase in serum LC-PUFA, stratified by the median hair mercury content (calculated based on model 2, right; data compiled based on Virtanen. 2012).
While there is certainly much that could be said about the overall study outcome, there are three things that are remarkable, novel and particularly noteworthy in the data in figure 3:
  • EPA is not only useless, without additional statistical shenanigan, it is even associated  (yet non-significantly) with an increased risk of CVD, when it's really high (+2% risk increase for each unit increase in EPA).
  • DHA is only protective, when the methylmercury levels are low (model 3 in figure 2 adjusts for that), when this is the case, however, each unit increase in DHA is associated with a whopping -19% decrease in
  • the statistical significance of the protective effects of DPA against sudden cardiac death is lost, when the data is adjusted for body mass index, pack-years of smoking and alcohol intake.
If we take the interactions with the hair (and thus presumably bodily) mercury load into consideration (see figure 3, right), it becomes obvious that hair mercury levels above the >1.28mg/g range renders both EPA and DHA practicually useless.

"Where do I get this DPA from; and what's that anyway?"

Figure 4: Enzymatic cascade from ALA to DHA; if you take a closer look the cascade does also explain why an increased conversion of ALA can competitively reduce the generation of EPA (see Portolesi. 2007)
Unfortunately, EPA and DHA are the two major forms of long-chain omega-3 fatty acids you will find in supplemental and dietary fish oil, so that your body will have to derive the DPA via Δ5-desaturase from EPA on its own (Leslie. 1985; see my illustration in figure 4 to get an idea of the whole cascade). This is not impossible, but obviously a rate limited step that could be avoided by direct supplementation, which is in fact something Miller et al. have done, only recently, and, as you have read, right here at the SuppVersity (see "On Short Notice" from July 29, 2012), which remarkable success (Miller. 2012).

Whether the beneficial effects of DPA are in fact related to its "reservoir function", Miller and his colleagues speculate about, cannot be said but would certainly constitute an intriguing research question for another rodent trial, maybe the mice in the Bourdineaud study would have been normal if they had had more DPA in their diets (see figure 1, right)

Bottom line: Until more scientific data is available (and probably still thereafter), there are actually three practical implications from this study you should bear in mind: (1) It does not make sense for anyone who carelessly shovels down tons of potentially mercury loaden fish to freak out about a tiny amalgam filling; (2) if you intend to benefit from the cardioprotective effects of fish oil, you better make sure that you are getting supplements and fish that have been tested for mercury, because the selenium alone obviously won't do the trick and save your ass... ah, pardon, your heart ;-) and (3) if you don't eat the worst offenders on a daily basis the benefits will probably still outweigh the negatives: I have recommended to fatty fish once or twice a week numerous times in previous articles and I don't see why these results would change anything about the recommendation.

References:
  • Bachmanov AA, Reed DR, Beauchamp GK, Tordoff MG. Food intake, water intake, and drinking spout side preference of 28 mouse strains. Behav Genet. 2002 Nov;32(6):435-43.
  • Bourdineaud JP, Marumoto M, Yasutake A, Fujimura M. Dietary mercury exposure resulted in behavioral differences in mice contaminated with fish-associated methylmercury compared to methylmercury chloride added to diet. J Biomed Biotechnol. 2012;2012:681016. Epub 2012 Jul 26.  
  • Calatayud M, Devesa V, Virseda JR, Barberá R, Montoro R, Vélez D. Mercury and selenium in fish and shellfish: Occurrence, bioaccessibility and uptake by Caco-2 cells. Food Chem Toxicol. 2012 Aug;50(8):2696-702. Epub 2012 May 22. 
  • Lemire M, Fillion M, Frenette B, Passos CJ, Guimarães JR, Barbosa F Jr, Mergler D. Selenium from dietary sources and motor functions in the Brazilian Amazon. Neurotoxicology. 2011 Dec;32(6):944-53.
  • Miller E, Kaur G, Larsen A, Loh SP, Linderborg K, Weisinger HS, Turchini GM, Cameron-Smith D, Sinclair AJ. A short-term n-3 DPA supplementation study in humans. Eur J Nutr. 2012 Jun 23.
  • Portolesi R, Powell BC, Gibson RA. Competition between 24:5n-3 and ALA for Delta 6 desaturase may limit the accumulation of DHA in HepG2 cell membranes. J Lipid Res. 2007 Jul;48(7):1592-8. 
  • Virtanen JK, Laukkanen JA, Mursu J, Voutilainen S, Tuomainen TP. Serum Long-Chain n-3 Polyunsaturated Fatty Acids, Mercury, and Risk of Sudden Cardiac Death in Men: A Prospective Population-Based Study. PLoS One. 2012;7(7):e41046.