"I was shocked to see that the EPA’s approach to distilling the scientific information obfuscated the evidence" – Prof Lianne Sheppard

EXCERPT: With glyphosate-based herbicides (GBHs) now ubiquitous in agriculture as well as schools, parks and home gardens, there is good reason to believe that most of us have been exposed to them. It’s possible that all of us, not just the most highly exposed workers, face some increased cancer risk from our exposure. It may be that glyphosate’s benefits outweigh its risks, particularly when compared to other weed control products. But determining whether that’s true is best decided by scientists using rigorous methods and evidence, not by flame-throwing in the mainstream media.
---

Glyphosate science is nuanced. Arguments about it on the internet? Not so much

Lianne Sheppard (Lianne Sheppard is a Professor at University of Washington's Depts of Environmental & Occupational Health Sciences, and Biostatistics)
Forbes, 20 Feb 2020
https://www.forbes.com/sites/thelabbench/2020/02/20/glyphosate-science-is-nuanced-arguments-about-it-on-the-internet-not-so-much/
[links to sources at this URL]

Just over three years ago, I was invited to join the U.S. EPA Scientific Advisory Panel (SAP) responsible for independently evaluating the Agency’s scientific review of glyphosate’s connection to cancer. I agreed, even though the time to prepare was short and my learning curve was steep.

I was intrigued both as a scientist, and as an at-home organic gardener, as I had understood that it is okay to use glyphosate - the key ingredient in Roundup, the world’s most widely used herbicide - on my property to treat a particularly pernicious weed problem, but I wasn’t sure of its safety. Not only would this service be an opportunity to find out more about glyphosate, it would also allow me to continue my longstanding commitment to use my statistical expertise to promote the public good. Little did I know the depths of the politically charged science I was stepping into.

In preparing for the December 2016 SAP meeting, I was shocked to see that the EPA’s approach to distilling the scientific information obfuscated the evidence. I found multiple and clear inconsistencies between the EPA’s Guidelines for Carcinogen Risk Assessment and its approach in its glyphosate review. For instance, in interpreting the animal toxicology studies, the EPA dismissed findings from studies that had evidence of a glyphosate response in one kind of statistical test but not in another. That’s despite the fact that its guidelines state that a statistically significant finding from either test is sufficient to conclude glyphosate is carcinogenic. It also dismissed evidence from the highest exposure groups in these studies, another apparent violation of its guidelines. This led me, together with two fellow SAP members Drs. Luoping Zhang and Emanuela Taioli, to dig deeper, ultimately publishing our own work on glyphosate.

The primary question we wanted to answer was this: “Does glyphosate cause cancer in humans?” To answer that question, we conducted a meta-analysis using the existing epidemiologic evidence. In contrast to the three previously published meta-analyses of glyphosate, we focused on the most susceptible exposure group in each study because this emphasis most directly addressed our question. The biology of cancer development suggests that higher exposures over longer time periods, particularly those exposures that occurred decades in the past, are more likely to result in cancer diagnoses if indeed glyphosate is a human carcinogen.

So in our analysis, we looked at the group in each study that had the highest cumulative exposure, the longest exposure duration, and/or the longest lag between exposure and cancer diagnosis. Of course, not every study defined these exposure levels in quite the same way, so we defined a set of rules in advance of our analysis to figure out the highest exposure group from each study. This approach helped us ensure we were being consistent as we analyzed the data across the different studies.

Our analysis estimated a 41% increased relative risk of non-Hodgkin’s lymphoma (NHL), with 95% confidence that this risk ranges between 13 to 75%, among those individuals with the highest levels of exposure to glyphosate. Which means they were 41% more likely to receive a NHL diagnosis than unexposed workers in their demographic. If a cause and effect can be established for this relationship, it means that glyphosate exposure causes NHL, at least for the most highly exposed workers. Showing such a causal relationship is extremely difficult in observational studies (such as those we used in our meta-analysis), but it is straightforward to conclude from randomized experiments. For example, evidence from animal and mechanistic studies shows that glyphosate exposure causes cancer. This convinced the International Agency for Cancer Research (IARC) in 2015 to conclude that glyphosate is a probable human carcinogen.

The increased risk we estimated may seem small, but it translates to an excess risk of 8 cancers per 1,000 among highly exposed workers. This risk exceeds the one in one thousand excess cancers that policy-makers generally consider acceptable for workers, and it far exceeds what is considered to be an acceptable risk for the general population (which ranges from one in ten thousand to one in a million). Furthermore, evidence of an increased risk of NHL from glyphosate exposure has broader implications. This is because cancer risks from substances that damage DNA tend to increase with increasing exposures, and scientists believe they lack a threshold below which there is no risk. This suggests that if glyphosate causes NHL, then members of the general population and workers who have much lower exposures may also have an increased NHL risk, albeit smaller due to their lower exposure.

This meta-analysis of the epidemiological evidence has received much attention, some of it negative. Some of our critics, including some academic scientists, claimed we cherry-picked the data, implying that we were searching for positive -- rather than objective -- results, and thus suggesting that we chose the data to report in our analysis after looking at the results.

Among scientists who leveled this criticism of our work are Dr. Geoffrey Kabat and Dr. Steven Salzberg in pieces originally published at Forbes. Dr. Kabat’s piece was removed for failure to meet Forbes’ editorial standards, and Dr. Salzberg’s piece referred to some of Dr. Kabat’s analysis without initially acknowledging said retraction (although the piece has since been updated to reflect this). Their arguments echo Bayer’s February 13, 2019 media statement that claimed our paper cherry-picked data.

Cherry-picking data to achieve particular results is never acceptable scientific practice, and in the case of our meta-analysis, this claim is not true. We focused on the most susceptible subgroup in each study in order to more demonstrably detect a possible link between glyphosate exposure and NHL if indeed such a link exists – by looking at the highest levels of exposure, we reduced the chance that other factors might be at play.

I have dedicated most of my career to the study of environmental exposures and their associated health effects, but this is the first time I have experienced other academic scientists questioning my work in the popular press. A hallmark of the scientific process is deep skepticism and independent peer review, and when researchers become concerned with some aspect of published findings, we’ve traditionally used mechanisms such as letters to the editor of the journal where the original results appeared, or publishing our own independent research on the topic. While all of my publications regarding glyphosate (here, here, and here) have stemmed from this process, neither Dr. Kabat nor Dr. Salzberg chose to follow this widely recognized path.

Even though scientific discussion is happening more often in the press and social media, and public debate about science and scientific integrity is healthy, I believe it is risky for scientists to pursue the truth in an arena that rewards inflammatory comment and lacks transparency about potential conflicts of interest. The pernicious effects of undisclosed conflicts of interest are particularly challenging when revenues are at stake, as they are with glyphosate. There is a long history of corporate influence in science directly and dangerously affecting public health through its influence on public policy.

With glyphosate-based herbicides (GBHs) now ubiquitous in agriculture as well as schools, parks and home gardens, there is good reason to believe that most of us have been exposed to them. It’s possible that all of us, not just the most highly exposed workers, face some increased cancer risk from our exposure. It may be that glyphosate’s benefits outweigh its risks, particularly when compared to other weed control products. But determining whether that’s true is best decided by scientists using rigorous methods and evidence, not by flame-throwing in the mainstream media. Public health is best served by policies to limit GBH exposures and to support additional research on their effects. Promoting messages such as Dr. Salzberg’s that “there is nothing to worry about” only serves the companies that sell them.