GMOs, Herbicides, and the New England Journal of Medicine

SprayerAn opinion piece was recently published in the New England Journal of Medicine, written by Dr. Philip Landrigan (an M.D.) and Dr. Charles Benbrook (PhD) titled “GMOs, Herbicides, and Public Health”. I was asked to review the piece this week, and I can only assume it will eventually make the rounds on social media. So I thought it was worth commenting here since it is related to herbicides and herbicide-resistant GMO crops.

The “Perspective” piece is basically a plea from Dr. Landrigan and Dr. Benbrook for “all aspects of the safety of biotechnology” to be “thoroughly reconsider[ed]”. However, in the two page opinion, they provide no evidence that crop biotechnology is harmful. In fact, Landrigan and Benbrook acknowledge that the National Academy of Sciences (NAS) “has twice reviewed the safety of GM crops” and they do not dispute the scientific consensus expressed by NAS that “GM crops pose no unique hazards to human health.” The way I read it, the entire Perspective piece seems to be a muddled conflation of two separate (albeit related) issues; the use of GMO crops and the use of herbicides.

Dr. Landrigan and Dr. Benbrook cite a list of concerns with two specific herbicides (glyphosate and 2,4-D), and use those concerns, among other things, to support a call for labeling of GMO foods at the end of the piece. But labeling GMO foods would provide little to no information about the herbicides (or other pesticides) used. For example, the herbicide 2,4-D can already be used on a vast number of crops that are not GMO (including corn, wheat, barley, and oats), so the GMO label would provide no indication about use of this herbicide. If the authors are truly concerned about the use of herbicides, their focus here on GMO crops seems misplaced. Some GMO crops significantly reduce pesticide use, so citing herbicide use concerns for a broad “GMO” label is counterproductive.

In the Perspective piece, Landrigan & Benbrook state that 0.4 million kg of glyphosate was used in the US in 1974 (the first year it was commercially available), and that use has increased 250-fold since then. This may sound concerning, but if you compare use of any popular product to the year it was first commercially available, the number will seem large. For example, iPhone sales have increased by a factor of 122 in just 7 years. In 1995, before the introduction of any glyphosate-resistant GMO crops, glyphosate use exceeded 12 million kg in the US (a 30-fold increase compared to 1974). Glyphosate use certainly increased dramatically in conjunction with adoption of Roundup Ready (glyphosate-resistant) crops. But even the 20-fold increase between 1995 to 2014 is not purely due to GMO crops, as other uses of glyphosate have expanded during that time due to cost decreases that followed patent expiration. Based on USGS data from 2012, approximately 18% of glyphosate was applied to areas that were not planted to GMO crops. This may seem nit-picky, but I do think it is worth noting since the “250-fold increase” can sound a little inflammatory out of context.

For all the nit-picking, it is worth noting that the 80% of 250 million pounds of glyphosate applied to GMO crops in the US is certainly a major selection pressure for glyphosate-resistant weeds. Dr. Landrigan and Dr. Benbrook cite glyphosate-resistant weeds as a primary reason why “fields must be now be treated with multiple herbicides,” but this point also deserves some context. In corn, for example, multiple herbicides have been a common practice since long before GMO crops were introduced. In the year 2000, before Roundup Ready GMO corn had gained widespread adoption in the US (and also before glyphosate-resistant weeds were growing in corn fields), corn growers were applying nearly three herbicide active ingredients per acre. The latest USDA data from 2014 show fewer than 3.5 active ingredients applied per acre. This suggests that while glyphosate-resistant weeds may certainly have increased the number of herbicides used per acre compared to 5 years ago, the change has been relatively modest when compared to herbicide use before adoption of GMO crops.

Dr. Landrigan and Dr. Benbrook also make several misleading statements in their opinion piece, including that the Enlist Duo registration decision by EPA “failed to consider ecologic impact, such as effects on the monarch butterfly and other pollinators.” The EPA provides all relevant regulatory decision information on its website, including the environmental risk assessment documents. Just one of the risk assessment documents for Enlist Duo contains over 100 pages of ‘ecologic impact’ data and interpretation, including environmental fate, degradation, aquatic and terrestrial organism toxicity and exposure estimates. The EPA also explicitly addresses how Enlist Duo will affect pollinators on an FAQ page dedicated to the Enlist Duo registration. It is inconceivable how Dr. Landrigan and Dr. Benbrook could conclude the EPA “failed to consider ecological impact” of Enlist Duo herbicide based on the vast amount of information provided in support of the registration decision.

Additionally, the Enlist Duo registration decision was not “made in haste” as Dr. Landrigan and Dr. Benbrook state. The Enlist Duo registration followed the US regulatory process in place for pesticide decisions, and even extended the public comment period before a decision was made. All public input was reviewed and EPA’s 30+ page response to those public comments is also publicly available on their website. It is worth noting that these two herbicides (glyphosate and 2,4-D) are two of the most widely used (and most widely studied) pesticides in the world, having recently gone through extensive reviews by several international regulatory bodies. There is really nothing about the Enlist Duo registration process that could be considered hasty.

Another misleading statement made by Dr. Landrigan and Dr. Benbrook is that the “risk assessment gave little consideration to potential health effects in infants and children, thus contravening federal pesticide law.” This claim was also addressed explicitly by EPA in their FAQ document. EPA concluded that after incorporating a 10X safety factor for children, and based on a “complete and very robust” data set, that the “risks were still acceptable for all age groups for all components of the assessment: dietary food and drinking water exposure, volatility, spray drift, residential, and aggregate assessment.” This claim is addressed in even more detail (600 words) in the EPA’s response to public comment, under the heading “B. Risk.”

Regardless of the headlines that will eventually accompany reporting on this opinion piece, there is nothing new presented here. And in fact, the information that is presented doesn’t really support the actions Dr. Landrigan and Dr. Benbrook are proposing.

Find this post helpful? Please consider taking a survey about science blog readership!


  1. Misinformation and innuendo seem to be the tools of choice for the anti-GMO group. Opinion pieces are just that – opinions. I hope people rebut the article with facts, as you have.

    1. The misinformation comes from the glyphosate manufacturer, which has waged a campaign of misinformation such as what is in this blog.

      But if you’re seriously taking the position that laboratory creations are identical to naturally occurring species – laboratory creations that could never occur in nature – then you probably also believe that plastic is the same as glass, pressboard is the same as wood, and movies are the same as books. Good 21st-century thinking.

    1. Alan, you are asking the question incorrectly. Different agro-chemicals have different effects. For example, a modest decrease in use of insecticides would greatly outweigh a very large increase in herbicide use. So it is vital to ask the question separately for insecticides and for herbicides. Even within the class of herbicides, they have very different toxicities. A five times increase in the use of the least toxic herbicide would easily be cancelled out if it produced a 1% decrease in the use of a commonly used herbicide that is many times more toxic.

      Asking the question about use of agri-chemicals by weight, the answer would that there has been an increase. Yet many studies have documented a substantial decrease in the environmental impact of agri-chemicals traceable to the use of GMO crops.

      1. I do largely agree with Charles. Weight is not the most useful measure, but it’s a rather complex qualitative question. How much harm is done by herbicides used today versus before HT crops? And, what’s the nature of the herbicides that are used with HT crops versus non-HT crops? One important point is that transgenics allows a different level of adaptation, by inserting foreign genes into an organism. In the case of Roundup Ready crops, it’s a bacterial gene encoding for EPSP synthase resistant to glyphosate to augment the plant’s natural EPSP synthase, which is inhibited severely by glyphosate. This means there is a new level of filtering possible, allowing the use of non-selective herbicides (which means that among non-GM plants there is nearly no resistance; of course, now, with time, that has changes as many weeds have evolved resistance). So, there may be less or more total weight of herbicide used, but the number and potency of molecules may be greater, potentially. Anyway, it’s apples and oranges and you really need to get “into the weeds” and study the details to arrive at meaningful conclusions.

      2. Transgenics does make possible a new level of herbicide tolerance that does make possible the use of extremely non-selective herbicides, and i think that was the point the authors were making. I don’t think they actually conflated herbicides with GM technology, but they explained their inter-relatedness in the current actual use. I think that is very valid.

        1. “Extremely” non-selective? A herbicide is either selective or non-selective. You can have varying levels of selectivity, but if it is non-selective, it’s non-selective. No need to try to make something sound scary by using hyperbolic adjectives. Just like there is no such thing as a “superweed” in relation to herbicide resistance. And GM crops are not “doused” with “massive” quantities of herbicides.

          1. Point here is that there is always *some* selectivity and we can also see that natural selectivity has been amplified by evolution of resistant weeds over time. But HT GMO crops allow an order of magnitude higher selectivity to be integrated in a crop/herbicide combo.

    2. I don’t know if you read Control Freaks regularly, but Mr. Kniss provided a pretty good perspective on that question at least with respect to corn here: The charts are very revealing, and actually don’t support a hypothesis that agricultural pesticide use in that crop is on an upward spiral.

      I don’t think it is unreasonable that the public focuses primarily on total usage as measured in weight and/or volume. I doubt few would dispute that lowering the amount of total usage should be an endpoint for science and agriculture to strive for. But I agree with Mr. Rader, that metric alone presents an incomplete, if not distorted, picture if the question is “Do current patterns and trends in agricultural pesticide use in the era of herbicide tolerant traits present greater environmental and food safety risk than pesticide use patterns preceding introduction of those crops.” Whether total amount of use is increasing or decreasing would only be one element of answering that question.

      1. Even if today’s pesticide/herbicide use is lower than it was before GMOs, we’d still be comparing two systems highly dependent on unnatural, synthetic and cancer promoting chemicals. We can’t justify today’s “conventional” farming with old “conventional” farming. Who would say that we shouldn’t transition from glyphosate-based farming to organic conservation tillage and no-till. That’s technology worth developing.

        1. As an oncologist, it’s disheartening to see young individuals with no established risk factors die of aggressive cancers. The incidence of cancer is on the rise. We’ve come a long way in terms of prolonging survival and understanding the molecular biology of cancer. The unfortunate reality that we grapple with everyday is that most cancers are so genomically deranged that a cure remains elusive for the vast majority of advanced cancers. The scary thing is that these derangements are dynamic and unstable. Cancer is continually evolving developing resistance to a variety of drugs we treat it with. It still continues to be the second leading cause of death in the U.S. ( more than half a million deaths per year)
          For instance microarray and genomic data demonstrate more than 500 different gene mutations within 2 different cancers arising from the same anatomic site. Difficulty with targeting these genes is that they also encode for normal physiologic functions. Hence our modern targeted therapy had added substantially to ones survival but at the cost of toxicity ( mouth sores, severe diarrhea, blistering in the hand and feet, heart damage, etc)
          The cost of some cancer therapeutics is around $100,000 per patient per year!
          Most times it does not result in a cure but prolonging life by a mere few months with toxicity from chemo/ biologics.
          We have to have respect for the genetic complexity of the human cell and the cancer cell. It’s not as simple as figuring out the molecular structure of the receptor and mixing the drug with the receptor to assess binding…
          I think the answer lies in prevention. Please, Let’s figure out ways to decrease the carcinogenic chemical inputs to our environment. I know I’m decreasing my job security but I’d much rather have my patients live healthier, more fulfilling lives than watch their lives get devastated by this terrible disease…

  2. Andrew, well done, thank you for this. I’ve been away (in the NW corner of your state without internet) and was hammered with questions about this piece. I was just finishing my rebuttal when I found this. You really should consider asking for equal space in NEJM. As a weed science expert, your ideas are important.

    Anti-GMO efforts know that physicians and dietitians are critical to manipulate. Opinion pieces like this have an influence because these professionals don’t have the time to dig into the research. Please consider writing something for them. Physicians should know your interpretations.

    1. Thanks for the note, Kevin. I’ve had a few people contact me to make similar requests. I’ll try to contact NEJM some time this week about a formal response. Hope you had a relaxing time in Yellowstone! -AK

      1. If you haven’t already written a formal reply to the NEJM piece, I urge you to do so. I ran across the piece while reading a completely unrelated article for an evidence review I’m conducting. As a nutritional biochemist who did postdocs in molecular bio and plant virology, who has been a study director at the NAS, who now works at a large thinktank and conducts evidence reviews for various federal agencies, and teaches a food policy class, my radar went up. My biggest concern is that the average MD and yes, the average nutritionist, have no perspective or knowledge with which to assess the veracity of anything Landrigan and Benbrook present: I myself am barely able to discern what I believe are the holes. And worse, Landrigan’s prominence lends great credibility to the piece Iin fact, I think he was chair of the committee that granted me my postdoctoral fellowship many years ago(:-)). So if you haven’t yet wriiten a reply to NEJM, please do and continue to spread the word!

  3. I thought the NEJM article was a very thoughtful appraisal. It certainly encourages members of the physician community to seek out scientifically valid evidence to some of the issues raised by Landrigan et al. The obvious concern is the carcinogenic potential for some of the herbicides and pesticides. Massive ubiquitous use of such is enabled via genetic modification
    The overall global incidence of cancer has been on the incline. AACR recently projected the incidence of ER + breast cancers to increase by 50% in 15 years!. Reasons for this increase are not completely clear. The role of “endocrine disruptors” such as Glyphosate have not been actively studied in relation to breast cancer risk in humans thus far. Limited mammalian cell line data and murine data certainly raise concern for a possible association.
    The data linking herbicides and pesticides to cancer in general are extensive. Most are animal and cell line data in addition to some recent human data indicating genotoxic injury in human DNA of workers exposed to Glyphosate. Large scale prospective human trials establishing disease causality remain impractical.
    Here is a very brief listing of some references providing links to the carcinogenic potential. The list is too lengthy to be a comprehensive compilation.


    – Glyphosate induces human breast cancer cells growth via estrogen receptors.
    Food chem Toxicology 2013; 2013 Sep;59:129-36
    – Long term toxicity of a Roundup herbicide and a Roundup-tolerant genetically modified maize. (Comment: This was critiqued due to some methodological flaws however adds to the evidence. Note : look at the author’s response to the comments. )Food chem Toxicology 2013 Mar;53:476-83
    – Major pesticides are more toxic to human cells than their declared active principles
    Biomed Res Int 2014;2014:179691
    – Glyophosphate: pathways to modern disease: Celiac sprue and Gluten intolerance. Interdis Toxicology 2013; Vol. 6(4): 159–184
    – Glyphosate induces human breast cancer cells growth via estrogen receptors.
    Food chem Toxicology 2013; 2013 Sep;59:129-36
    – Glyphosate-induced stiffening of HaCaT keratinocytes, a Peak Force Tapping study on living cells. (Comment: Contradicts the claim that Roundup is not absorbs by human keratinocytes)
    J of structural Biol 2012 Apr;178(1):1-7
    – Alteration of estrogen-regulated gene expression in human cells induced by the agricultural and horticultural herbicide glyphosate. Human Exp Toxicol. Sep;26(9):747-52
    – Biomonitoring of genotoxic risk in agricultural workers from five colombian regions: association to occupational exposure to glyphosate. (Comment: Data for genotoxic damage in humans with occupational exposures to Glyphosate; transient and mild genotoxic damage however it contradicts the current claim that Round-up is not absorbed or metabolized by humans)
    J of Toxicol Environmental Health 2009;72(15-16):986-97

    1. GKahn, Thanks for stopping by to comment.
      1) With respect to your comment of “endocrine disruptors such as glyphosate”. There isn’t evidence that glyphosate acts as an endocrine disruptor. Based on extensive testing, the EPA recently found “no convincing evidence of potential interaction with the estrogen, androgen or thyroid pathways.” (Source:
      2) For your request to provide “SCIENTIFICALLY VALID PUBLISHED DIRECT HUMAN EVIDENCE”, please see a summary of the human evidence of carcinogenic potential of glyphosate here:

  4. Thanks for sharing the references.

    I looked at them. Obviously this represents an independent appraisal by the EPA. It by no means establishes or refutes causality of herbicides to cancer.

    There were some methodologic issues that merit further investigation.

    Most carcinogens are unlikely to cause cancer based on acute exposures. For instance if one smokes 10 cigarettes a day for a 2 weeks it is unlikely to cause cancer versus years of chronic, long term low dose exposure. Radon is also a widely established carcinogen however many of us have it in our basements and don’t die on acute exposures. It causes no detectable animal toxicity acutely at the doses typically found. I can cite several other examples…
    Hence it brings to question what the long term follow –up was in the animal models tested.

    Also here are some other issues

    – In the ERTA assay the negative control failed to evoke a response; hence the assay is typically considered invalid; however the committee continued to maintain that there was no association with endocrine disruption
    – In the FSTR assay significant reductions were noted (upto 55%) in the vitellogenin levels. Although it did not have any immediately detectable clinical or pathologic consequence it does not exclude long term effects. As humans we are lucky to be much more physiologically complex than our aquatic counterparts. I can’t tell you how many studies we come across in cancer therapeutics that demonstrate excellent in vitro and in vivo efficacy in cell lines and animal models however human clinical trials are negative.
    – It would be more informative to have a study looking at pharmocokinetic assessments of Glyphosate metabolites in the serum/urine to understand its role in normal physiology and consequently any aberrations from normalcy
    – I looked at the “human evidence” that was provided. It represents the “EPA’s opinion” on safety. Although concern was raised from metanalysis about possible associations with NHL, the values were not statistically significant. Not reaching statistical significance is often times a numbers game. If the study is adequately powered the likelihood of attaining a significant p value or odd ratio is higher.
    – With regards to the case control studies: you correctly point out that only 3% of workers were exposed to glyphosate. This in and of itself renders the validity of data questionable. We can’t make meaningful interpretation where only about 3% of the study population was exposed to the agent in question

    Bottomline, the human data that are provided are certainly thought provoking. I agree that they do not establish causality however they no not refute it either. I would have liked to see a prospective trial in human agricultural workers exposed to Round up ( and other pesticides) with a detailed analysis of pharmacokinetic parameters and pharmacodynamic parameters such as genotoxic damage — ideally with long term follow up data on incidence of cancer or even precancerous lesions

    1. I’m surprised that in all this commentary and the article, nobody has mentioned the review level article that establishes a positive association of a 2.0 risk factor between glyphosate exposure and B-cell lymphoma.

      Schinasi, L (April 23, 2014). “Non-Hodgkin lymphoma and occupational exposure to agricultural pesticide chemical groups and active ingredients: a systematic review and meta-analysis.”. International Journal of Environmental Research and Public Health 11 (4): 4449. doi:10.3390/ijerph110404449. PMID 24762670. Retrieved 31 August 2015.

  5. There’s research that states that glyphosate binds(chelates) to soil nutrients and harms microbial balance in soil. Glyphosate is documented highly estrogenic and promotes endocrine disruption, which is linked to cancer. Instead of bickering over which conventional herbicides and pesticides are the worse, let’s discuss how we can decrease dependence on unnatural, harsh chemicals, improve the soil and farm with natural inputs.

    1. But this article is also about GMOs and GMO policy. The proponents of GMOs claim that GMOs are safe.

      But how could they possibly make such a statement? ‘GMO’ refers to thousands of bioengineered products that have already been manufactured as well as any that may be manufactured in the future. The already manufactured GMOs haven’t been tested for safety; it is unlikely that GMOs still to be developed will be either, in the absence of a strict regulatory requirement to do so..

      Would you be comfortable with a pharmaceutical manufacturer’s claim that all pharmaceuticals that will be developed in the future will safe? That’s what’s being asked of you with regard to future laboratory-engineered food.

  6. Pingback: SciBabe

Comments are closed