Sunday, February 15, 2015

Unsafe at Any Dose?

In this post we'll see why we no longer have to take any notice of claims that GMO crops and herbicides are not causing us harm.  But first, let me put this in the context of this blog's major goal: to explain the importance of Swanson et al. and to defend it against its critics (for reference see below).

Yes, it’s high time to up-armor for the coming fight.  Ripples are already spreading across the blogosphere, but so far the only counterargument I’ve seen that went beyond the hoary old “bad journal”, “bad credentials” Monsantoite b.s. was that the authors should have separated different areas of America for statistical purposes. That’s because, if they were right, surely there should be more chronic diseases where there was heavy spraying and less where there was little or none.
GMO advocates never tire of repeating the mantra “The poison is the dose”—even things you eat unthinkingly every day, like table salt, can poison or even kill you if you eat too much.  Just the other day, a comment by Chris Preston on a post on the Biofortified page (  claimed “Toxicity always depends on the dose. Regulators address whether a product can be used in such a way that the dose received is below a level which may result in toxicity.”  In technical terms, all substances with toxic potential are monotonic—their toxicity may be zero at very low doses, but increases proportionately with the size of dose.

This looks like a general law, not just one for toxic substances. Surely the more you consume of anything, the greater the effect?  The more food you eat, the fatter you get.   The more alcohol you drink, the drunker you get.  Your common sense and your senses tell you that.  They also tell you that the sun goes round the earth (not vice versa), that continents can’t creep around the globe (they’re solid, lifeless rock, goddammit!), that we couldn’t possibly be kissing cousins to chimps.  But if our senses and our common sense told us how things REALLY worked, there wouldn’t be much need for science, would there?  What science does is prove that the counterintuitive thing is all too often the truth, and I haven’t even mentioned quantum mechanics.

While we’re talking about science, let me emphasize one of the most important things about it.  It moves on.  It’s always moving on.   It’s not like religion, where you must believe exactly what people believed hundreds or even thousands of years ago, or else be branded as a heretic.  If you truly believe in science, you must always be ready to change, because science is always changing, and once it’s changed, what was science yesterday isn’t science any more.  “The poison is the dose” is a case in point.

The assumption behind “The poison is the dose” is that damage from any toxic substances can be avoided if you simply make sure that people don’t get too much of it.  And the mechanics of that seem straightforward enough.  There’s a nice summary at  “New medicines or chemicals which may affect the health of humans are required by law to be tested on animals…Safety tests begin with acute toxicity testing, where the animals are given a single dose of the test compound. The aim of the tests is to determine the range between the dose that causes no adverse effect and the dose that is life-threatening.”

 Alas for that.  According to Vandenberg et al. (“Bisphenol-A and the Great Divide: A Review of Controversies in the Field of Endocrine Disruption” Endocrine Reviews 30.1. 75–95, 2009), “a safe dose determined from high doses does not guarantee safety at lower, untested doses that may be closer to current human exposures.”
Why not?  It’s because for any toxic substance you can plot a response curve, with a strong effect at or near the top of the curve and a weak or null effect at the bottom.  And there is not just one possible curve--here's a sample of several:


Note that in all the figures, low-dose is to the left of the graph, high-dose to the right. The left-hand A and B graphs are the kinds of curve once thought to be universal (and still are by pro-GMOers).  The right-hand C curve—the U-shaped curve—is very different, and probably the hardest one for GMO supporters to deal with.

That’s because of the mode of testing described above—start high, work down until effects aren’t apparent, leave what you think is a wide enough margin and announce a safe dose.  In other words, you plot only the right-hand side of the U-curve.  There’s no way you could find that, at still lower levels, harmful effects could begin again (left-hand side of the graph).  But that's exactly what the graph means.  It means that if a toxic substance has a U-curve but you like a good Monsantoite assume it's monotonic, it may very well have serious consequences that you literally cannot know about till after they've happened.

By now I’m sure GMO defenders will be saying. “Well, what’s the so-called scientific evidence for all this?  Some rubbish published in a pay-for-play journal with a 0.something impact factor, I’ll bet.”  Well, sorry, guys.  Endocrine Reviews has the highest Impact Factor ranking of the 89 journals in the ISI category of endocrinology and metabolism. Of the total 5,684 surveyed by ISI, EDRV's Impact Factor ranking is #20.”  (source: ResearchGate, but you can also consult the original ISI lists.) The journal’s impact factor is 19.36, and the paper itself has been cited in 537 other journal articles and books.  We’re not talking junk science now; we’re talking Gold Standard in Endocrinology.

Move on another three years, and the same journal publishes “Hormones and Endocrine-Disrupting Chemicals: Low-Dose Effects and Nonmonotonic Dose Responses” (Vandenberg et al., Endocrine Reviews 33.3. 378-455 (2012)).   This is right in our ball-park because it specifically includes glyphosate in its list of non-monotonic dose-response curve substances that cause substantive harm:

Chemicals by chemical class      Nonmonotonic effect                       Cell type

Glyphosate-based herbicide        Cell death, aromatase activity       HepG2 liver
(Round-Up)                                 ERβ activity                                   cells

Note that this paper has 564 citations, even better than the previous one, especially since they cover only a two-year period.  It includes nearly 850 citations of supporting work.

Fallback position for GMOers:  “This stuff’s very controversial, you’re cherry-picking data, good science says the opposite”.  Well, six years ago (quite a while in science at nowadays speeds) Laura Vandenberg wrote:  “Although scientific inquiry is a dynamic give-and-take among researchers with different opinions and viewpoints, the so-called controversies surrounding low-dose effects and NMDR curves should be put to rest, given that they now affect public health decisions [My italics, DB].  These phenomena have been demonstrated time and again for a sufficient number of endocrine-related endpoints, and they no longer merit being considered ‘controversial’ topics.”  In other words, this is the new orthodoxy in toxicology.

So what has all this got to do with Swanson et al.?  Well, first and foremost, it gets them off the “why no data by area” hook.  If glyphosate has a low-dose effect, then there is no reason to expect people in high-spraying areas to have more chronic diseases, and therefore no point in separating data from different states or regions.  

But the work on response curves goes much further than that.  As Vandenberg et al. point out at the end of their 2012 paper, “The concept of nonmonotonicity is an essential one for the field of environmental health science because when NMDRCs occur, the effects of low doses cannot be predicted by the effects observed at high doses.”  This means that when GMO advocates tell us that low doses of glyphosate are harmless, their claims no longer have any valid scientific evidence.  To the contrary, the knowledge that glyphosate is non-monotonic and an endocrine disruptor makes it all the more probable that it does cause substantive harm.   What’s the next step towards proving this?  Well, how did people first find evidence that tobacco caused lung cancer?  Through epidemiological studies and correlations!
The tobacco comparison is a story that deserves its own post, and will get it.  For now, it’s sufficient to note that this is the importance of Swanson et al.  Through epidemiology, the paper builds a prima facie case for supposing that glyphosate could indeed cause the rise of certain chronic disease conditions in America.  Given that we now know what toxicology can and can’t prove, Swanson et al.’s claims can no longer be dismissed with blanket denials--they must be further investigated.  And however that investigation turns out, we’ll still be able to tell GMO defenders that their claims of pesticide safety aren’t worth the paper they’re written on.

Reference Genetically engineered crops, glyphosate and the deterioration of health in the United States of America” by Nancy L. Swanson, Andre Leu, Jon Abrahamson and Bradley Wallet,  Journal of Organic Systems, 9(2), 2014 (


  1. Some interesting concepts here. I scanned through the Swanson et al paper, but I won't pretend I understood it all. However, it seems clear that correlations exist. As you pointed out, causation remains to be proven, but correlation is normally enough to justify further research. I think it will be interesting to see this moved through the rest of the scientific process.

    As a side note, I'm looking forward to seeing EPA's position on NMDR curves as they relate to endocrine disruptors. I know a lot of people don't put much faith in government bureaucracies, and they certainly move slower than molasses in January, but they certainly ensure the thoroughness of their evaluation.

    1. No problem, Kent, I'll walk you through it in the course of the next few posts. Glad you accept it demands further looking into--that's all the authors claim! Yes, I too would like to see EPA's position on it, but have they said they are going to take one? Not to the best of my knowledge. You're right about molasses. Supposed to be a National Academy of Sciences report on GMOs etc. actually scheduled--for 2016!

  2. Hi Derek,

    That glyphosate is likely to be harmful even at low doses is a big part of why the findings of a study like this one are so important.

    That led me to cite one of the same studies you cite here in my most recent news summary.

    These studies and the state of the science they establish for endocrine disruptor toxicology would be sufficient to blow out of the water even non-corrupt EPA tolerance levels. Of course EPA limits are tendentious anyway and meaningless other than as a gauge of how much poison the corporations expect to sell, since the EPA mechanically raises the levels in response to corporate petitions.


  3. Wow. Do you see where this argument leads?
    The author makes the case that the effects-to-dosage toxicity of substances can be non-linear; and that in cases we even find "U"-shaped curves, where toxicity is high at high dosages, low at medium dosages, but then high again at low doses.

    Note that while he's ostensibly talking about glyphosate toxicity, this could be equally true for all substances; and why not consider this when evaluating benefits as well as toxicity?

    With this view of things, how can we assess anything with confidence? How can we decide what doses are low, medium and high? When do we decide we trust a substance to be safe (or beneficial), when someone can always argue that a dosage lower than what we evaluated might prove to be highly toxic (or have completely different benefits) than our spectrum of tests revealed?
    Assuming there are substances with toxicity and/or benefits that behave this way, we need a whole new paradigm for evaluating effects. Simply saying we can't trust safety or benefits thresholds set using a linear model is in invitation to total paranoia. If our old model can't be trusted, its only fair to ask for a new model that can.
    Until a new methodology is put forward, all this argument does is mess with our heads.

    1. "we need a whole new paradigm for evaluating effects."
      That's exactly what Vandenberg et al. say in their article. That's just a fact of life. I wish I could find such a methodology, but that's way above my pay grade, so I'll leave it to those who can--and will!