Transgenerational Epigenetic Inheritance and Glyphosate: Conclusions

By Alma Laney and Alison Bernstein

This post is the sixth and final post in a series about transgenerational inheritance, epigenetics, and glyphosate that address questions raised by the publication of the paper, Assessment of Glyphosate Induced Epigenetic Transgenerational Inheritance of Pathologies and Sperm Epimutations: Generational Toxicology.

Do the reported results support the conclusions from the authors?

In the conclusions section of the article, the authors had this to say:

Glyphosate exposure of the F0 and F1 generation had negligible toxicity and pathology, which supports direct exposure having low risk, however, the transgenerational germline mediated inheritance promotes significant pathology and disease.

Unfortunately for the authors, it is difficult to draw these conclusions from the data presented. The experimental design issues may have doomed the experiment before it began. This may be more of an issue of inadequate reporting and the experimental design may actually be fine – we just cannot tell from looking at this paper.

Even if the experimental design is actually appropriate, the other issues we have described make it very hard to conclude that the observed pathology is a result of glyphosate-induced transgenerational epigenetic inheritance and not another confounding effect that is independent of glyphosate. As to whether this is relevant to humans and real-life exposures, the dose is clearly too high to have any real relevance to real-life scenarios.

Reporting from both “sides” has not accurately explained this study

In a completely predictable manner, reports on this study have ranged from citing this study as proof positive that glyphosate is dangerous and should be completely eliminated from use, such as the recent releases from Moms Across America, to the knee-jerk reaction of the skeptic community with invalid criticisms of the problems with the study and complete dismissal of the idea of epigenetic inheritance.

The question of how epigenetic changes are inherited across generations in complicated and very unclear at this point. Even with one of us (Alison) working in the field of developmental neurotoxicity and epigenetics, it took a lot of time to dig into the methods and results to understand what was going on here.

Alma’s experience with this paper is informative and a useful reflection on the importance of recognizing the limitations of our expertise.

I myself fell into this trap despite my training including coursework on epigenetics, albeit it was epigenetics in microbes, which is different than what happens in plants and animals. When I first read through the study, I too noted the Venn diagram and how there weren’t shared epigenetic changes across all three generations and I thought I had identified a critical error. Inherited changes need to be inherited, right?
In discussing this paper with Alison, I quickly realized how little I knew about how epigenetics works in mammals. I then started reading quite a few papers on epigenetic inheritance. I was so far out of my area of expertise that I had to spend almost as much time looking up definitions and asking Alison questions as I did actually reading the papers and reviews that she suggested. In all of this, I barely scratched the surface enough to question if I would even be able to address a paper like this.
I could see that there was some things wrong with the experiment, but what I could identify were issues with statistical analysis and with control groups being removed from consideration. But these issues were just the tip of the iceberg. Simply put, despite being highly educated, I was no expert on this topic. Alison, on the other hand, looks at epigenetics and epigenetic inheritance in her research. She very much qualifies as an expert on this topic.
If it were up to me to write this article alone, you’d just have a short article about stats and control groups rather than the much more in-depth analysis we presented. Expertise matters. There’s no way to get around this. With studies like the one we addressed here, having an expert to break down what it actually means is incredibly important. If even scientists within the field can have questions about a study like this, how can we expect reporters and even scientists in other fields to do a better job of communicating the findings?

A study like the one we are discussing here highlights why expertise is so important. Even people with expertise and training in other areas can have difficulty interpreting a study like this, especially when the methods aren’t clearly stated in the article as was the case here. If people with extensive training can misinterpret this type of study easily, then the reporting will most likely misinterpret the study too.

Do we need to think of the grandchildren?

This question is an important one. Are we able to show there is a concern based off of the data presented in this paper? Unfortunately, the methodological and reporting issues make it nearly impossible to draw any conclusions about whether transgenerational inheritance is occurring or if this is mediated by epigenetics. In addition, the chosen dose is largely irrelevant to actual human exposures levels and the method for introducing that dosage is not a way that people would normally be exposed to glyphosate. This really makes it hard to say that the normal levels and routes of exposure would result in harm.

A study with more precise methods and analyses would be needed to sort out if the findings of this study are realistic or not. But even if we took the findings at face value and uncritically accepted the authors’ conclusions, this wouldn’t necessarily mean we need to worry about poisoning future generations.

Transgenerational effects are, by nature, reversible and modifiable. So as an individual living in a world of complex exposures (both protective and risk factors), when one considers the entire risk landscape, the behaviors and exposures that mitigate risk will still mitigate risks of prior and future exposures for current and future generations.

View the other parts of our series on transgenerational epigenetic inheritance:

Transgenerational Epigenetic Inheritance and Glyphosate

By Alison Bernstein and Alma Laney

The paper Assessment of Glyphosate Induced Epigenetic Transgenerational Inheritance of Pathologies and Sperm Epimutations: Generational Toxicology reported transgenerational epigenetic inheritance and increased disease rates after glyphosate exposure. Not surprisingly, the paper generated a lot of attention and discussion. Due to the focus on glyphosate by activist groups and recent lawsuits, we’ve taken an in-depth look at the state of the science on transgenerational epigenetic inheritance, the data in this paper, and the larger body of work from this lab.

Whether glyphosate exposure causes health problems through transgenerational epigenetic inheritance is an important research question. The original EPA reference dose is based on a transgenerational phenotype, even though this result has been determined to “be spurious and unrelated to treatment since more extensive evaluations in subsequent reproduction studies conducted at much higher doses did not replicate the offspring effects” (as explained in the draft human health assessment for glyphosate).

In this series, we address questions about transgenerational inheritance and epigenetics in general, and this glyphosate study in particular.

Credibility is our currency: conflicts of interest and research misconduct

We need to talk about conflicts of interest (COIs) in scientific research. Specifically, we need to talk about the difference between COIs and research misconduct. There seems to be a misunderstanding in the media and public conversations that a COI is research misconduct. While a COI may lead a researcher to commit research misconduct, a COI is not, on its own, research misconduct.
What are conflicts of interest?
A COI is a situation in which a person has multiple competing interests, financial or other, that have the potential to compromise or bias their judgment or objectivity. COIs exist whether or not decisions are affected. COIs merely recognize the potential for wrongdoing based on conflicting motivations.
COIs are generally divided into two categories: intangible and tangible. Intangible COIs involve academic activities and scholarship, while tangible COIs involve financial relationships. Tangible COIs can include intellectual property rights, consulting fees, honoraria, gifts, ownership or royalties. Examples of intangible COIs are: delaying publication of a manuscript to benefit the next grant application, or a researcher’s bias in interpreting data towards his or her own hypothesis.
Different types of COIs have different potentials that can lead to bias. A free pen or a free sandwich provided at a seminar is not the same as funding a study, which is also not the same as covering a portion, or all, of a scientist’s salary.
What is scientific misconduct?
In the United States, the US Office of Science and Technology Policy has defined research misconduct as follows:
Research misconduct means fabrication, falsification, or plagiarism in proposing, performing, or reviewing research, or in reporting research results.

Fabrication is making up data or results and recording or reporting them.
Falsification is manipulating research materials, equipment, or processes, or changing or omitting data or results such that the research is not accurately represented in the research record.
Plagiarism is the appropriation of another person’s ideas, processes, results, or words without giving appropriate credit.
Research misconduct does not include honest error or differences of opinion.

In science, everything comes down to credibility. Research misconduct erodes trust between colleagues, between scientists and funding agencies, and between the institution of science and the public. Thus, research misconduct is taken very seriously within the scientific community. Consequences of research misconduct can range from retraction of papers to being banned from receiving funding, depending on the severity and scope of the misconduct. Being found guilty of research misconduct often marks the end of a scientist’s career, as seen by a severe decline in number of publications and funding after the misconduct.
What’s the difference between a conflict of interest and research misconduct?

While conflicts of interest may lead to research misconduct, they are not evidence of misconduct nor are conflicts of interest necessarily misconduct on their own. The presence of a COI may demand closer scrutiny of the research to determine if misconduct or bias affected the interpretation of the results. However, a COI itself is not research misconduct, nor does the existence of a COI automatically mean that research misconduct occurred. This is not to minimize the importance of the disclosure of COIs. It is this very transparency that allows us to identify problems, limit COIs and scrutinize research that may be biased. In science, credibility is our currency. Transparency and disclosure about conflicts of interest are critical to maintain our credibility.
Disclosing COIs makes us all better able to identify our own and others’ biases. This is why all US scientists at federally funded institutions disclose COIs and external activities. Each institution has an office that is devoted to reviewing COIs and implementing measures to help minimize these conflicts and their effect on research outcomes. Disclosure of COIs is taken very seriously. In fact, failure to disclose COIs is itself misconduct.
Disclosure is also important beyond the official institutional disclosure system. Typically, when scientists give presentations in our own university, at other institutions or at conferences, speakers include relevant COIs at the beginning and acknowledge funding sources at the end. Journals also have requirements about COI disclosure that must be met before publication. Failure to do so can result in retraction of a paper.
It’s also important to note that the presence of COIs, and even misconduct, does not necessarily negate the data. Determining if misconduct or bias occurred and if it invalidates the data is complicated and requires close scrutiny of the data and the specific situation.
A few examples from the archives of Retraction Watch help to illustrate how complex it is to sort out issues of COI, misconduct and data validity.

In this example, authors forged the paperwork to add an author who did not contribute. This is a clear case of misconduct that resulted in a retraction, but the data is still valid.
Research misconduct involving methodological flaws, fabricating data or other invalid manipulations of data would invalidate the data collected. In this case, a collaborator falsified data resulting in two retractions. This clearly invalidates the published data.
Failure to disclose a COI is misconduct, but whether a paper is retracted or corrected depends on whether there are findings of additional misconduct. The examples listed on this link show the variety of ways that the failure to disclose COIs are handled.
It’s also possible for there to be methodological flaws resulting from honest mistakes and errors that lead to retraction of a paper, but these are not considered misconduct.

There is often criticism that scientists do not take conflicts of interest seriously enough to appease the concerns of the lay public. However, this argument conflates COIs with research misconduct. The scientific community requires disclosure of COIs to help identify and monitor possible cases of misconduct. Awareness and disclosure of COIs allow the public and the scientific community to assess whether misconduct or bias occurred. Scientists are not generally penalized for the mere presence of COIs; they are penalized for research misconduct.
We need to stop automatically punishing scientists for having COIs. It’s not having a COI that’s a problem, it’s how COIs influence research that can become problematic. Let’s make sure we are judging the behavior of scientists, not based on just the existence of COIs, but based on how they act in the face of those conflicts.
This article originally appeared on The Sound of Science blog, and was published on the Biofortified Blog on 10 March 2016. At that time, we believed it was important to revisit a discussion of research ethics and to clarify what constitutes research misconduct and what does not due to the allegations of research fraud and misconduct raised against Dr. Federico Infascelli. We are republishing it in 2018 because these are still important concepts that both scientists and the public need to understand.