Smoke & mirrors in the pesticide approvals process
For Laurin and Livia
www.kremayr-scheriau.at
eISBN 978-3-218-01100-6
2017 by Verlag Kremayr & Scheriau GmbH & Co. KG, Wien
All rights reserved
Image Part II Chapter 1: courtesy of Andreas Rummel
Image Part II Chapter 4: courtesy of GLOBAL 2000
Translation: Patricia Lorenz and Joanna Scudamore-Trezek
Project management & copy-editing: Sonja Franzke, vielseitig.co.at
Cover design & letterpress: Silvia Wahrstätter, buchgestaltung.at
Prologue: The wrong WHO
Part One USA 1973 to 1991
Chapter 1: Suspected fraud
Chapter 2: Cancer alarm
Chapter 3: Miraculously exonerated
Chapter 4: A licence to kill
Part Two Europe 2012 to 2017
Chapter 1: System failure
Chapter 2: Mobilisation
Chapter 3: Smoke and mirrors
Chapter 4: Entanglements
Chapter 5: Cracks in the system
Epilogue The future
Appendix
Glossary
Endnotes
Thanks
16 May 2016. My instructions were to refrain from working, just for once. Simply to spend a stress-free afternoon with my daughter at the playground. Nothing unusual for a public holiday in spring, but it wasn’t to turn out like that. While my daughter clambered over the climbing frame on this particular Whit Monday, I typed the term glyphosate into Google News on my mobile phone.
The force of habit.
For I work as a biochemist for the Austrian environmental organisation GLOBAL 2000. The main focus of my work is problematic commonplace chemicals and how to protect people from the dangers and risks they present. Over the past year I had been working with a group of environmental activists, consumer protectionists and scientists from various European countries towards our shared goal of preventing the European Union from renewing its approval of a carcinogenic pesticide. In only two days‘ time, in Brussels, the 28 EU Member States would vote on the issue which had kept my European colleagues and myself busy for months. Would glyphosate receive renewed approval in the European Union despite the WHO classifying it as “probably carcinogenic for humans”?
What Google News returned in response to my query from the playground was unexpected bad news; I read the words: WHO gives thumbs up for glyphosate. Germany’s daily newspaper FAZ wrote: New evaluation of risk: WHO does not believe glyphosate causes cancer1. Even Die Zeit carried the headline: WHO scientists classify glyphosate as not carcinogenic.2
I was shocked. At a stroke this volte-face would make all our work of the recent months obsolete. How could this be? And above all, why should the WHO’s cancer research agency (IARC) have suddenly revised their assessment?
Then came the revelation – someone else was behind the retraction: the joint FAO/WHO meeting on Pesticide Residues (JMPR). This scarcely pronounceable name describes an international pesticide advisory panel of rather dubious repute. However, at this moment the “retraction” of the JMPR threatened to tip the scales at the forthcoming EU vote: in just two days‘ time glyphosate could be given renewed approval. The champagne would then be flowing at Monsanto! What was little known, however, was that the JMPR had already once given a highly dubious all-clear to Monsanto’s glyphosate back in the 1980s. Even then the JMPR contradicted another expert panel which the previous year had classified glyphosate as “possibly carcinogenic in humans“.
Back in the 1980s it was the American environmental protection agency, the U.S. EPA; today it is the WHO’s cancer research agency IARC. But who or what is this JMPR?
While the WHO’s cancer research agency IARC exclusively uses studies which are open to public examination when classifying cancer, the JMPR primarily relies on studies which are carried out by the pesticide manufacturers themselves and kept under wraps. Where the IARC strives for independence and transparency in its decision-making, many see the JMPR as dominated by industry and opaque in its dealings. While the IARC assesses whether a chemical by its very nature is carcinogenic, irrespective of dosage and method of intake, the JMPR examines only the risk posed by pesticide residues in foodstuffs.
One of the first to welcome the JMPR’s positive assessment on Twitter was the Executive Vice President and CTO of Monsanto, Robb Fraley. Only a year earlier Fraley had ranted and raved against the IARC. Now he happily tweeted: WHO/FAO refutes WHO/IARC; glyphosate safety re-affirmed! But there was another tweet on the same topic: French journalist Stéphane Foucart of Le Monde revealed an interrelationship which throws a different light on the JMPR’s judgement: Alan Boobis and Angelo Moretto (ILSI associates) were not only members of the JMPR WG on glyphosate, they were its chair and co-chair. Someone had done some careful research, because Alan Boobis and Angelo Moretto are two scientists whose names regularly crop up whenever there are attempts to turn industry-friendly recommendations into regulations. Both have important roles at the International Life Science Institute (ILSI), a global science institute largely financed by major food and chemical corporations including Monsanto, Bayer, Nestle, BASF, and many others.3
Despite this, however, many newspapers appeared the next day falsely proclaiming that the WHO gives the all-clear for glyphosate. And the timing of this purported exoneration of the pesticide, just one day before the decisive vote, was naturally devastating, both for us and all those wanting to block glyphosate.
The trip to the playground was abruptly ended by this news. My confidence that our work would bear fruit in terms of the EU decision was seriously dented. Now it was essential to rectify the basic facts with a statement to the press. Firstly, that the WHO had not made a “retraction”. Secondly, that there are still unresolved conflicts of interest within the JMPR. And thirdly, that the JMPR is exclusively concerned with pesticide residues in foodstuffs. Too many topics to squeeze into a 1-page press release. For in the end this was only the latest development in a story which had started in the USA in the 1970s. To explain all this would fill a book. This book.
On 4 April 1983, the US Federal Court of Chicago opened a trial that was to become one of the most extensive trials in the history of Illinois. By the time the sentence was pronounced in October 1983, its records filled more than 16,000 pages.1 The legal authorities had been investigating the case for six years. Newspaper commentaries had called it the biggest case of scientific fraud in the history of the USA, maybe even in the history of the world. The fraud was suspected to have been committed by Industrial Bio-Test Laboratories (IBT Laboratories), formerly the largest private testing institute for industrial chemicals in the United States. In the dock were Joseph C. Calandra, the founder and former owner of IBT Laboratories, as well as three of his former leading scientists. They were charged with routinely falsifying test reports on substances in drugs, personal care products and pesticides and thus to have wilfully deceived the US-American regulators.
The fraud had already been uncovered back in 1976. Adrian Gross, an attentive investigator at the US Food and Drug Administration (FDA), the authority responsible for the authorisation of drugs and food additives, had become suspicious when examining a study on a painkiller carried out by IBT Laboratories. The test data seemed “too good to be true”.2 None of the rats had developed tumours, Adrian Gross explained later, although every pathologist knows that rats and mice in long-time studies also develop spontaneous tumours, and show a certain degree of mortality. However, according to the IBT study, all of the animals were clean.
During a more precise examination of the data of this “Naproxen study”, Gross discovered oddities in the data tables: rats which had been listed as dead in one section of the study suddenly reappeared in another section of the study. “We do not really believe in animals coming back to life after death”, Adrian Gross noted in his report. Due to the appalling sanitary and hygienic conditions in the animals’ housings, IBT Laboratories indeed struggled with a mortality rate among the laboratory animals that, in many cases, prevented them from doing tests in accordance with the regulations. In order to hush up this fact, IBT technicians were ordered to “replace” perished animals with healthy ones. When the advanced decay of the dead animals made an autopsy impossible, this was documented internally with the abbreviation TBD / TDA, “too badly decomposed/technician destroyed animal.” 3 This probably explained the riddle of the lack of tumours.
The Naproxen painkiller study was only one of 22,000 studies that the IBT had carried out for manufacturers and authorities during their three decades of existence. Hundreds of chemicals that had since become part of thousands of daily products had been classified as safe because of the test results. One of these chemicals was to receive particular attention in the forthcoming trial: Monsanto’s antibacterial agent triclocarban (TCC). What the interrogation of witnesses was to unearth about the safety testing on TCC and the role that Monsanto played in this is suggestive of how far the company was ready to go to push through its interests.
TCC, an anti-bacterial additive for soaps, had attracted the attention of the FDA at the beginning of the 1970s. There had been indications that TCC had caused deformities in the testicles of rats fed with the anti-bacterial agent. This presented Monsanto with a problem: at around the same time, the chemical company had requested an increase in the permitted levels of TCC in bar and liquid soaps. Monsanto therefore asked the private IBT Laboratories to do a feeding study on rats. The study was supposed to convince the authorities that TCC was safe. Interestingly, at about the same time, Monsanto toxicologist Paul Wright left his job at Monsanto to take up a position at IBT Laboratories. There, as section head for rat toxicology, he supervised the very study that was to prove the safety of TCC. However, as early as half way through the 24 month feeding study, it became apparent that the laboratory rats resisted this (putative) plan. In June 1972, pathologist Donovan L. Gordon discovered testicular lesions in the male animals. He noted his observations in an interim report, which caused unexpected reactions.
The numerous attempts that Gordon’s superiors and Monsanto representatives made between 1972 and 1976 to convince the pathologist that the damages he identified were not caused by TCC, but had to be due to other reasons, were given a lot of space in the trial. In a report that is as readable as it is disturbing, US-American journalist Keith Schneider outlines the details. It was published in the Spring Edition 1983 of the Amicus Journal by the environmental organisation Natural Resources Defense Council, and taken up in numerous US-American media. Here is an overview of the most important events as they relate to Monsanto’s TCC:4
•On 11 October 1972, before the interim presentation of the study to an FDA panel, two Monsanto scientists impress upon Gordon how important it is to present TCC in a good light. Seemingly with success, for the minutes do not mention testicular lesions. However, Gordon continues to think that the testicular lesions are caused by TCC.
•6 October 1973: three months after the end of the feeding study, Paul Wright, who by then has quit IBT and is again working for Monsanto, meets the head of IBT, Joseph C. Calandra. He is concerned that Gordon’s interim report continues to describe testicular lesions that have occurred with large and small doses of TCC. He pushes Calandra to convince Gordon to change his conclusions, because factors such as stress, gender, age and nutrition could also be responsible for the animals’ lesions.
•On 22 January and 21 February 1975, Gordon’s findings in the TCC study are criticised at two further meetings at IBT.
•25 August 1975: Gordon, who sticks to his conclusion, has to re-examine and discuss the tissue samples with Dr William Ribelin, a pathologist hired by Monsanto. Gordon is able to convince Monsanto pathologist Ribelin that his observations indeed prove a treatment-related effect in all three dose groups5. Later, he will testify in court that Ribelin also gave him a handwritten copy of his report, in which Monsanto was informed that Ribelin had also arrived at the conclusion that TCC caused testicular lesions in all three dose groups.
•In January 1976, a last unsuccessful attempt is made to persuade Gordon to change his findings. Then Calandra asks an unexpected question: had there been any significant decomposition in the rats he studied? Gordon confirms that this was the case in some rats but not with others. Calandra then informs Donovan L. Gordon that he will remove his findings from the report and replace them with the wording that decomposition precluded meaningful evaluation of the testicular tissues.
•11 May 1976: Monsanto delivers the final TCC report, in which rats in the low dose group no longer show testicular lesions, to the FDA. The latter sees its concerns about testicular atrophy due to TCC dispelled, and recommends an increase of the admissible levels of TCC in bar and liquid soaps.
•April to July 1976: IBT is suspected of fraud. The TCC study is one of the first to come to the attention of the investigators. Monsanto has to repeat the study and commissions the private Bio/dynamics testing laboratory to do this. IBT, meanwhile, shreds hundreds of letters between IBT and its clients that, investigators presume, could have proven that large chemical companies had known about the IBT fraud.
•June 1982: a Monsanto spokesman explains to the magazine Mother Jones6 that the “hopelessly flawed IBT study” had been replaced by a study by Bio/dynamics Laboratories. According to the spokesman, that study shows that a human would have to eat 26 bars of soap a day for a normal lifetime in order to consume the amount of TCC that caused significant toxic effects in the test animals. Monsanto’s bacterial killer TCC is considered vindicated.
•21 October 1983: Paul Wright and two of his co-accused are found guilty by the jury of key charges (IBT founder Joseph C. Calandra was able to evade the trial because of health problems). Wright is jailed for six months and loses his job at Monsanto.
The man who had uncovered the scandal, Adrian Gross, had been asked one year before the start of the trial why corporations would allow their products to be subjected to phony or inadequate testing. Did the companies not want to know if their product causes cancer, reduces fertility or damages the DNA? And did they not risk being sued and financially destroyed by marketing dangerous products? Gross answered with a counter question: “How are you going to prove that one pesticide caused cancer in somebody? First of all… a person, unlike a rat, is exposed to hundreds if not thousands of them.” Chemical companies, Mother Jones stated, know that they are fairly safe from individual product liability lawsuits. In the rare event that they lose a case, damage awards tend to be relatively small when weighed against the profits generated by the product.”7
The further course of the history of TCC proves him right: Monsanto escapes further investigations. This is not changed by the fact that, according to Mother Jones, the new contracting laboratory of Monsanto, which in the end vindicated TCC, also came to the attention of a U.S. EPA/FDA inspection shortly afterwards because of inadmissible practices and deficits. TCC successfully survived on the US market. According to a study by the U.S. EPA, more than 80% of deodorant soaps in the USA contained TCC at the turn of the millennium. In 2008, however, a team of researchers at the University of California found that TCC is an endocrine disrupting substance (see box on page 22) and causes changes in male sexual organs in tests with rats.8
A worrying result for, in 2014, a study sponsored by the National Institute of Health showed that 80% of the pregnant women examined excreted TCC in their urine and a fifth of the newborns had the bacterial killer in their blood at the moment of their birth.9 The authors of this study expressed their concerns about the endocrine-disrupting potential of TCC. On 6 September 2016, the FDA finally decided to ban the use of TCC in bar and liquid soaps, with the exception of medical purposes. More than four decades after the pathologist Donovan L. Gordon noted the first clear indications of this in his study report, the FDA thus stated that TCC “could pose health risks, such as (…) hormonal effects”10.
Endocrine Disrupting Chemicals, EDC
The WHO links endocrine-disrupting chemicals to a worldwide increase in fertility disorders, such as a decrease in the quality of sperm or deformities of male sexual organs, an increase in hormone-related forms of cancer, such as breast, prostate or testicular cancer, as well as a series of other diseases of modern civilisation, such as diabetes, cardiovascular diseases, etc. Their damaging effect is due to the fact that, because of their (coincidental) structural similarity to hormones, they can disrupt sensitive hormonal control processes in the organism. Like hormones, they can sometimes take effect in very low doses. The organism is most sensitive to the interference of endocrine-disrupting chemicals during so-called critical developmental windows, which is why foetuses, infants and adolescents are particularly sensitive to these substances. Decades can pass between the exposure to endocrine-disrupting chemicals and the manifestation of an illness due to this exposure.11
We are the lab rats!
In the US media, the IBT fraud case caused some venomous and strong reactions. In its edition of 4 July 1982, the Tallahassee Democrat asked:
Feel safe? … and then adds: Don’t. They’re testing products on you.12
One of the most comprehensive, radical and simultaneously unerring analyses was offered by the monthly Mother Jones magazine13 in its June and the following July edition of 1982. In part 1 of its analysis14 the editors preceded their report with a letter that said:
A year-long team investigation has found that our entire regulatory structure is so corporate-dominated, so riddled with incompetence and corruption that it almost completely fails to protect the American people.
The paper bluntly uncovered the inherent conflicts of interest that contracting laboratories, such as IBT, face:
Private testing laboratories live and die by the positive results generated on products they test for safety. Your future as a laboratory owner or administrator is in serious jeopardy if you find too many products unsafe.
The magazine however, was hardest on regulatory authorities:
The illusion of safety is not always created by corrupt testing laboratories or deceptive corporations. Too often, regulatory agencies look the other way or blindly accept the conclusions of obviously self-interested research.
The total failure of the regulatory system was closely linked to the concern for health. The authors of Mother Jones presented concrete figures on the incidence of deformities in newborns and of annual cancers, and linked them to the disastrous failure of safety checks by authorities.
More than 800,000 new cases of cancer are expected this year, and, according to the Center for Disease Control, nearly three out of every 100 children born in 1982 will have “serious birth defects”. Some of this carnage is inevitable. It is caused by combinations of human error, bad luck and “acts of God”. But far too much of it is caused by products we place our trust in because they are labelled safe.
Suddenly, a question that nobody had been interested in for years was being debated in the national papers: is it smart to leave the examination of sensitive chemicals to those who produce and sell these chemicals? And why are the test reports kept a secret?
Lessons learned…
Will those responsible now draw the evident and necessary conclusions from the scandal? Will the registration system be fundamentally reformed? Unfortunately, the answer is no. The essence of the registration system remains untouched, although a lack of transparency, inherent conflicts of interests and a lack of control were identified as inacceptable weak points. The manufacturers are still allowed to test the health risks of their products themselves or outsource the tests to a laboratory of their choice. They are involved in the creation of the test report, write the corresponding dossier and suggest to the authority how it should interpret the findings. The only thing that really changed after the IBT fraud was the regulatory requirements for (private) testing laboratories. The Good Laboratory Practice Act (GLP), adopted by the US Congress in 1978, defined a binding framework for carrying out toxicological studies where these are used for the registration of a product by the authorities. The GLP lays down clear standards for the housing and treatment of laboratory animals, the storage of chemicals and the full, daily documentation of results and test details. Compliance with these standards was to be ensured by a team of inspectors created for this very purpose. The crass falsification of studies, the invention of data – something that was part of daily life at IBT – was to disappear with the new laboratory standard. The times when studies were “too good to be true” (Adrian Gross) were over. From now on, the correct estimation of risks and dangers would be much less frequently jeopardised by falsification of test data than by incorrect (statistical) evaluation and an improper interpretation of the results by testing institutes, manufacturers and authorities.
But there was a second, pressing problem to be solved. 22,000 IBT studies were probably not even worth the paper they were written on. The authorities assumed that about half of them had served for the registration of products for the American market. Of those, in turn, a significant proportion concerned the registration of pesticides.15 The U.S. EPA thus had no choice but to test at least the most important of these innumerable pesticide studies – one after another – for their plausibility. Studies that proved to be unreliable had to be repeated by the manufacturers. The 200 active substances in pesticides whose registration was mainly based on IBT studies also encompassed a herbicide that Monsanto had put on the market only a few years earlier, in 1974: glyphosate. The two cancer studies – one with rats and one with mice – that Monsanto submitted to the U.S. EPA at the beginning of the 1970s to prove that glyphosate was not carcinogenic had been performed in the IBT laboratories. They were now suspected to be useless or maybe even manipulated.
Glyphosate – from pipe cleaner to record pesticide
N-(Phosphonomethyl)glycine, better known as glyphosate, is a broad-spectrum herbicide. It non-selectively kills all plants by disrupting a central metabolic pathway that is common to all plants, as well as bacteria and fungi.16 The fact that this pathway does not exist in humans and animals has often been used as an argument for the alleged safety of glyphosate. Glyphosate is the biggest-selling and most-used active substance in herbicides. In 2012, around 90 chemical companies in 20 countries produced 720,000 tonnes of glyphosate.17 Monsanto’s Roundup was the first weed killer with glyphosate as an active substance.
Initially, herbicides containing glyphosate were only used to keep fields free from weeds before sowing. In the 1980s farmers also began to kill off cereals that were ready for harvest with glyphosate, in order to speed up the drying process and to control the time of the harvest. In the 1990s, the US corporation Monsanto succeeded in making plants resistant to glyphosate with the help of genetic engineering, and patented them.
This enabled farmers to use Monsanto’s Roundup also after sowing and during the entire growth period of the plants. Monsanto’s patent on glyphosate ended in 2000.
Today, there are dozens of herbicides containing glyphosate on the market. Besides their use in farming, they are also used in public spaces, around railway tracks and in private gardens.
Glyphosate was discovered by Swiss chemist Henri Martin, who was the first to synthesise the chemical in 1950. At first, glyphosate attracted little interest. In 1964, Stauffer Chemical Company patented glyphosate as a “chelator”, because it binds minerals, such as calcium, magnesium, manganese, copper and zinc. Glyphosate was first used as a pipe cleaner to remove calcium deposits from pipes in hot water systems.18 The US corporation Monsanto also studied glyphosate as a potential water-softening agent. During these studies, it discovered its herbicidal effect and patented glyphosate as a herbicide in 1969. In 1974, Monsanto introduced the herbicide Roundup, which contained glyphosate, to the US market.
On 21 August 1978 – two years after the IBT scandal and five years after the introduction of glyphosate to the market – U.S. EPA toxicologist William Dykstra had to investigate whether the IBT-cancer studies reliably proved that glyphosate did not cause any tumours in rodents. Dykstra subjected the old Monsanto studies to the same process that Adrian Gross had used two years before to uncover the deficiencies of the IBT painkiller study: he took a closer look at the data tables. In his report19, Dykstra stated that, because of the loss of 70 animals from the control and test groups and the low number of animals examined histopathologically1, he could not firmly conclude whether glyphosate had an oncogenic (i.e. carcinogenic) potential or not. In total, there were 66 animals whose date of death could only be estimated based on the date on which they disappeared from the individual weight tables. The cancer study with mice offered a similar picture: “too many mice are missing”, records William Dykstra. That same evening the U.S. EPA decides that Monsanto has to repeat the rat and mouse cancer studies.20
William Dykstra, U.S.-EPA-Memorandum from 21. August 197821
One of the consequences of the IBT scandal was, therefore, that the EPA had to recognise, five years after the first registration of glyphosate, that the two cancer studies submitted by Monsanto were unusable. But why only then, when it was so obvious that not even all of the animals had been examined for tumours? On what basis had the U.S. EPA accepted the two studies at the time? Had it maybe not even looked at the original studies that could not be accessed by independent scientists and that could be checked solely by the authorities?
About thirty years later, this is exactly what a European registration authority would have to admit. For the moment, however, let us remain in the USA of the 1970s.
1A histopathological examination is the microscopic examination of tissue samples.
Monsanto has to repeat its two cancer studies with rats and mice. This is not good news. On the one hand, a feeding study costs a lot of money. However, that might be the smaller problem. It would be much worse if the results did not match. In other words: if the rats in the test groups – i.e. the animals who are administered glyphosate in the feed – were to develop tumours significantly more often than the rats in the control group. In this case Monsanto would have a real problem because that would mean that glyphosate might be considered carcinogenic and then the Delaney rule would apply.
The Delaney Rule1, also called the Delaney Clause, is a 1958 amendment of the Federal Food, Drug, and Cosmetic Act, named after James J. Delaney, a Member of the US Congress. It states that chemicals that cause cancer in humans or animal testing may not be used in foods.
From the consumer point of view this sounds like a good idea, but it was anathema to industry from its very beginning. The latter mainly criticised the Delaney Rule for not distinguishing between highly carcinogenic and mildly carcinogenic substances. Industry still claims that it is “unscientific” not to take into account the effective risk with proper use.
The decision of the U.S. EPA that both IBT cancer studies were unreliable thus really put Monsanto’s glyphosate to the test. If the pesticide were shown to be carcinogenic, its use in agriculture (if at all) would only be possible with restrictions. However, big money cannot be made with pesticides that can be sprayed only along roads and railway tracks.
Towards the end of 1978, Monsanto commissioned the first replication study to determine the risk of cancer caused by glyphosate: a 26-month feeding study with rats.
In 1981, the moment had come: the rat study is finished and is impressively titled:
A Lifetime Feeding Study of Glyphosate (Roundup Technical) in Rats (Report by GR Lankas and GK Hogan from Bio/dynamics for Monsanto. Project #77-2062, 1981: MRID 00093879) – Including the study’s 4-volume Quality Control evaluation of the Bio/dynamic assessment performed by Experimental Pathology Laboratories, Inc. (2,914 pp).
That same year Monsanto files an application for the nondisclosure of the study. The corporation regards the study as its own property and a trade secret. Since the U.S. EPA obviously shares this view, the study is henceforth kept under wraps. From this moment, the regulatory authority is the only authority in the world that has the opportunity to control the correctness and plausibility of the data and conclusions in this study. However, this is the very thing that the U.S. EPA is not going to do.
The study is assessed on 9 February 1982 by U.S. EPA toxicologist Dykstra (the very man who, four years previously, had declared the IBT cancer studies with glyphosate to be useless). The assessment report2 delivered by Dykstra was not to become his masterpiece. Instead, it offered a taste of things to come in terms of the twists and turns that European authorities would also take, three decades later, in sweeping the cancer findings from feeding studies with glyphosate under the carpet.
One of Dykstra’s serious mistakes occurs at the very beginning, when he “overlooks” two of the three statistically significant tumour findings recognised today (pancreatic, thyroid and testicular cancer). Dykstra indeed declares the incidence of testicular tumours in the treated male animals to be the only noticeable tumour findings. The distribution of the testicular tumours is presented as follows in the U.S. EPA memorandum:
William Dykstra, U.S. EPA Memorandum from 18 February 19823
This means that none of the 50 rats in the non-treated control group had developed testicular tumours.
However, in the low-dose group three animals had done so, in the mid-dose group there was one, and in the high-dose group there were six.
Testicular tumours (male animals)
Monsanto, 1981
Cancer study with rats
The decisive question when assessing such a cancer study is: can the present distribution of tumours be attributed to the tested substance (here, glyphosate) or did it occur by chance? In other words, how great is the probability that the distribution of tumours that developed is treatment-related? We speak of “probability” because coincidence is always an important factor and/or possible disruptive factor. Tumours, indeed, can also occur spontaneously. So if those animals who were administered the tested substance manifestly developed tumours more frequently than the untreated animals from the control group, such as in the present rat study by Monsanto, statistical tests are used to test how probable it would be for this event, or a more extreme event, to occur if the administration of glyphosate would not have any influence on the result (i.e. if it is assumed that glyphosate is not causing cancer). This probability is defined as significance level “p”, or probability value. One convention that has been defined with these kinds of cancer studies is that findings are seen as statistically significant if the probability value is lower than 5 percent (p<0.05)2. Further valuable leads for the assessment of tumour findings can also be delivered by comparing present data with data from untreated control groups of other, comparable feeding studies, in so-called historical controls.
If one and the same cancer study is evaluated in contradictory ways by different bodies, this is quite often a result of the incorrect application of statistical tests and/or historical controls. Glyphosate is a prime example of this. This is why it is worth taking a closer look at these important tools and the basic rules for their correct application:
The approach to the statistical evaluation of cancer studies with rodents has basically remained unchanged for the past 40 years. It is now described in detail in various OECD reports and guidance documents4. Basically, two statistical test methods are used: the “trend test” and the “pairwise comparison test”. Study findings are considered to be statistically relevant if one or both statistical tests, i.e. the trend test or the pairwise comparison test, show a probability value of less than five percent (p<0.05).5
The trend test takes into consideration the entire curve progression (e.g., in the present case 0/50, 3/50, 1/50, 6/50) and tests whether a significant, dose-dependent trend of an increase in the incidence of tumours can be identified in the total results. Since the trend test takes into consideration all four test groups together (a control group plus three dose groups), it has a higher “statistical power” and is the more sensitive of the two tests – provided that the effect shows a linear (proportional) dose-related response. Since this is not the case in the present cancer study, the trend test does not show a significant result.
The “pairwise comparison”, in contrast to the trend test, only ever considers two groups at a time and compares them: in the present case, the incidence of tumours in the high-dose group (6/50) is compared to the incidence of the control group (0/50). The test questions whether this difference is statistically significant and the answer is yes: p = 0.013. The probability value is thus significantly lower than five percent.
Historical controls can deliver further valuable indications for the assessment of study findings. These controls are made with tumour findings from the control groups of comparable feeding studies.
They should provide information on a “normal”, spontaneous incidence of tumours, in addition to the concurrent control (usually 50 animals at most). To ensure that the use of historical controls results in meaningful and consistent conclusions, it is vitally important to observe certain rules that are also laid down in the current OECD recommendations. In order to ensure that the results are comparable, historical controls have to come from the same rat or mouse strain and must have been completed under the same test conditions, i.e. same test duration within a comparable window of time (a maximum of five years previously), ideally with the same pathologist and in the same laboratory. In order to prevent distortions due to erratic samples (outliers), using the median value from all control groups instead of the mean values is recommended. It is fundamental to stress that the concurrent control group is always the most important consideration for each evaluation.6 Statistical tests (and if necessary also historical controls) have to be used in line with the OECD recommendations. This is a prerequisite for correctly assessing whether tumour findings in feeding studies with rodents are conclusive or not.
Equipped with this knowledge, we will now take a look at William Dykstra’s approach when assessing the tumour findings above. In doing so, we will discover a pattern that we will encounter quite often later on. That is, regulatory authorities – be it in the USA or in Europe – completely reject significant cancer findings with glyphosate as a chance occurrence on every occasion. It almost seems as though the regulatory authorities have developed a method of their own that enables them to ignore indications and evidence within a (seemingly) ordered and seemingly scientific system.
Dykstra now has to answer the following question: are the tumour findings of the present feeding studies treatment-related? Let us have a look at how Dykstra, in four steps, arrives at a conclusion that contradicts the facts:
•Step 1: William Dykstra controls whether the most important prerequisite for a treatment-related effect, i.e. the statistical significance of the findings, is fulfilled and states in his assessment dated 9 February 1982 that this is the case:7 in a pairwise comparison with the control group (0/50), six rats with tumours in the high-dose group (6/50) are statistically significant (p=0.013).
•Step 2: Dykstra compares the incidence of tumours in the high-dose group with the incidence in the historical controls. Dykstra finds that the 12% incidence of testicular tumours in the high-dose group is higher than the incidence in control groups from other, comparable feeding studies (mean: 4.5%, maximum: 7%) and writes:
Historical controls thus also suggest a treatment-related effect.
•Step 3: Instead of drawing the conclusion that is suggested by the findings above, Dykstra now refers to a scientifically doubtful “interpretation” from the study report of the contracting laboratory of Monsanto:
•Step 4: Following this dubious argumentation (what does he mean by saying that the significance of the 12% incidence in the high-dose group is “not known”? What does he mean by saying they may represent a “biological variation”?) the investigator declares the present findings as not treatment-related.
Thus according to William Dykstra the carcinogenic potential of glyphosate is negative. Monsanto’s feeding study is stamped “top secret” and shelved in the filing cabinet at the U.S. EPA where it will remain for a year.
The overlooked tumour findings
In February 1983, a phone call by the Canadian regulatory agency sets the ball rolling again, and puts William Dykstra in an unpleasant position. His Canadian colleague wants to know how the U.S. EPA interprets the significance of malignant thyroid cancer in female rats. The catch is that Dykstra, apparently, has not even noticed malignant thyroid tumours up to now. His assessment stated that, with the exception of testicular tumours, no significant differences could be found between treated and untreated rats.
Now Dykstra has to have another, closer look at Monsanto’s rat study. The U.S. EPA memorandum dated 15 February 19838 states that the tumours of the thyroid gland are not only significant but cause for concern. In addition to the benign thyroid tumours that have a high incidence in this strain of rats (Fig. p. 35 left), six animals in the high-dose groups have developed much rarer thyroid carcinomas. They have done so in a significant and dose-dependent increase (fig. right). In the control group, however, only one rat developed a thyroid carcinoma.
The result is also alarming for the following reason: an arbitrary, non-treatment-related incidence of thyroid carcinomas in 6 of 47 animals (13%) is, statistically speaking, highly improbable. The “historical controls” of the testing institute indeed show that only 5 of 512 untreated female animals (1%) developed spontaneous thyroid carcinomas.9
The opposite is true for the benign tumours (fig. above left). The “pairwise comparison” does not deliver any significant difference to the control group for any of the test groups, and the “trend test” does not result in a significant, dose-dependent trend either. Fittingly, the average incidence of benign tumours in the historical controls by Bio/dynamics (9%) is nearly precisely of the same order as the average incidence of tumours in the study over all four test groups (8.8%). Therefore, benign thyroid tumours can doubtlessly be seen as not treatment-related. They form the normal background incidence of benign tumours of the thyroid in this strain of rats.
The U.S. EPA is now faced with a dilemma. The dose-dependent increase of malignant thyroid carcinomas really suggests only one conclusion: the effect results from the treatment with glyphosate. If the authority recognises that, it has to revoke its assessment dated 9 February 1982, with all the resulting, regulatory consequences. Moreover, the authority would have to admit that it had overlooked important data during its first examination of Monsanto’s rat study. That is something nobody is keen on doing. And an authority that had just had to put up with massive media criticism because it had not been looking closely enough at the faked IBT studies over the years was even less likely to be keen on doing it.
However, there is a tempting suggestion as to how William Dykstra and his authority can get out of this dilemma. The suggestion comes from somebody who must have an even more vital interest in keeping the EPA assessment as it stands: the manufacturer of the pesticide. Let us now have a look at the proposal that was written by the Senior Product Toxicology Specialist of Monsanto, Rick B. Oleson, in an addendum to the rat study:
Rick B. Oleson: An Addendum to a lifetime Feeding Study of Glyphosate in Rats10
Add the benign and the malignant tumours together, and the problematic significance of the thyroid carcinomas disappears – simply vanishing into thin air. The overall result can now be presented as a chance result. A tempting proposal: easy to implement and scientifically unacceptable. However, for the U.S. EPA in 1993, this seems to be a welcome gift enabling them to conclude, against the obvious facts, that there are no positive cancer findings for the thyroid glands either. And this is exactly what the U.S. EPA is doing now. It agrees with the argumentation put forward by the Monsanto toxicologist and rejects the significant thyroid carcinomas as not treatment-related. Before doing so, it requested and received confirmation from three “external pathologists” that they would also support this approach. However, at least two of them have – it seems11 – business connections with Monsanto. In any case, the thyroid carcinomas have been ticked off as “not treatment-related” and Monsanto’s rat study can disappear again into the filing cabinet.
It would be interesting to know why Dykstra had not seen the result on the tumours of the pancreas. This was the third significant tumour result whose existence we know of today from Monsanto’s cancer study with rats. As there was apparently no friendly authority who called the EPA to inform them of the existence of these significant tumour findings this time, the authority did not have to come up with a means of getting the significant findings out of the way.
That these findings became known at all, even though the study had been classified, is due to the fact that they suddenly appear in the second assessment of glyphosate by the U.S. EPA in 199112 (without an apparent explanation as to why they had been overlooked up to then). They are described in enough detail there for the International Agency for Research on Cancer of the WHO (IARC/WHO) to take these findings into account when assessing glyphosate in 2015.
