Is Glyphosate Probably Carcinogenic? Probably Not! 


In May 2015 anti-GMO lobbyist and ignorance profiters were handed down some glorious news. The World Health Organisation (WHO) had just declared the Glyphosate was probably carcinogenic. Finally validation that GMO’s are evil and monsanto is trying to rape mother earth for every last cent. I read this with some surprise as I was under the impression that there had been numerous safety reports etc… showing the glyphosate (the key ingredient in roundup) was relatively safe. A body within the WHO called the International agency for Research on Cancer (IARC) was the devision responsible for this new claim. Aside from sounding like something James Bond would be sent to destroy or otherwise compromise IARC stated aim is ‘to promote international collaboration in cancer research.’ I feel I should state that I actually have a lot of respect for the WHO. I am an avid supporter of their vaccination work and principles of education in helping improve human health, it is with this in mind that I write this. What I am about to say I do not do lightly, the level of scholarship depicted in the IARC report is on par with Andrew Wakefiled of MMR/Autism fame.

In 2015 they released a short communication of their upcoming full monograph’s on the likely cancer causing properties of 4 separate pesticides. Tetrachlorvinphos, Parathion, Malathion, Diazinon, and Glyphosate (1). In this short press release communication the authors laid out several damming conclusions regarding the safety of Glyphosate. They stated that there was sufficient evidence in animals that it was carcinogenic and limited but potentially positive evidence of carcinogenicity in humans. Reportedly linked to Non Hodgkin’s lymphoma (NHL). I first read this shortly after it came out and was immediately annoyed by the lack of citations for almost all of the claims. A total of 6 citations were given for numerous claims over the two paragraphs. 3 of these citations were for a single claim. This set my researcher senses tingling… why didn’t they cite the primary literature? One key citation they had included was a 2006 report conducted early by the WHO in 2006 (2). This was easy enough to track down and within moments I had my jaw on the floor! However instead of writing this then, I held back. I had to wait until the full monograph was released otherwise I could jump the gun and find the reported did contain some evidence that was released after this earlier report. But now the full monograph (3) is out in the public and I am still shocked.

Let us start with the Animal work, as this to me is the most breath takingly dishonest (intentionally or not). The section of the short report can be split into three separate claims.

Claim 1. ‘In male CD-1 mice, glyphosate induced a positive trend in the incidence of a rare tumour, renal tubule carcinoma.’

The claim was also the second thing that set my researcher senses off, the phrase ‘positive trend‘. Scientist choose their words very carefully, if something has a definable response we tend to say it is significant, normally followed by just how significant and the test we used to look at it. Positive trend to me is a weasel word. If you squint your eyes enough and allow your bias to come through you can see the data in such a way as to support your hypothesis. A bit of background on the study; 100 mice (50 and 50 male female) we given different doses of Glyphosate in water and allowed to drink it as required for 24 months. The EPA did this to study the safety of Glyphosate (4 & 5). They found that 5 of the mice developed a form of kidney cancer. The IARC states that this is a rare type of cancer and they made a note of by it going on to state that in historical checks of CD-1 mice from another study only 1 of the 725 mice had been seen to develop this kind of cancer.  There is a slight problem with all of this, one of those cancer infected mice was in the control group. There also seems to be confusion about the actual existence of the cancer as the pathologists and histopathologist disagreed to the numbers of infected mice. Further the statistical test used even comes with a warning from the original lab that ran the stats stating

‘This linear trend test often gives incorrect results’

So let’s look at the full monograph, here’s what they said in the longer report,

‘The Working Group considered that this second evaluation indicated a significant increase in the incidence of rare tumours, with a dose-related trend, which could be attributed to glyphosate. Chandra & Frith (1994) reported that only 1 out of 725 [0.14%] CD-1 male mice in their historical database had developed renal cell tumours (one carcinoma)’

However this ignores that a rare cancer appeared in the control line and statistically analysis saying test is often wrong. There is a serious question that needs to be addressed here, why are the authors of the monograph happy to ignore the appearance of a rare cancer in the control group but happy to attribute the cancer in the tests to the glyphosate, all the while ignoring the bad stats and the unconfirmed existence of the cancers! Further adding back in the control line tumour means that the results are no longer significant. It also ignores another report submitted to the WHO in 1993 by Atkinson et al which did not show any similar effect, using the same mice and similar doses, although it was submitted to the WHO it is not a publicly available study therefore it was not counted. Even though the WHO used it previously to determine the safety of Glyphosate. The 2006 WHO report summed up Atkinson’s study as such.

‘In conclusion, administration of glyphosate to CD-1 mice for 104 weeks produced no signs of carcinogenic potential at any dose.’

There is also the problem with the citations used by IARC, it includes a 1986 EPA report which contradicts the earlier assessment (6). This report makes it certain that the authors feel that the earlier 1985 EPA study was reporting tumour levels consistent with expected control levels and that the histopathology was indicative of spontaneous growths and not due to the added compounds (Glyphosate). Further; that in fact the stats used are incorrect and when correctly applied there is no significant findings. Did IARC even read this citation?

The 1986 EPA report is as damning throughout of the study and conclusions (6)

The lancet article also states, along a similar line, ‘A second study reported a positive trend for haemangiosarcoma in male mice’

This is one of the cited statement (the 2006 WHO report), but there is that weasel word again ‘positive trend’. So let us see what the citation has to say about this study.

‘Owing to the lack of a dose–response relationship, the lack of statistical significance and the fact that the incidences recorded in this study fell within the historical ranges for controls, these changes are not considered to be caused by administration of glyphosate.’

So again let me translate for you, a non-significant (potentially caused by random chance), non dose-responsing (adding more glyphosate didn’t have a consistent effect) and a within historical controls (these mice are known to have levels of this disease naturally) study being reported as evidence that glyphosate can potentially cause cancer. It seems the lancet is happy to publish the opinions of researchers as I think it is fair to say they haven’t provided any science yet.

Claim 2. ‘Glyphosate increased pancreatic islet-cell adenoma in male rats in two studies.’

Here we have a similar story to claim 1, The levels of adenoma was within normal range for those mice nor were the results significant and the final nail: they did not progress into cancer. The IARC monograph had said this of the studies.

‘The range for historical controls for pancreatic islet cell adenoma reported in males at this laboratory was 1.8–8.5%. The Working Group noted that there was no statistically significant positive trend in the incidence of these tumours, and no apparent progression to carcinoma.’

Yet the press release cited these results as evidence for potential harm. Further from 2006 WHO report which also examined this paper.

‘The incidences of this lesion were 1 out of 58 (2%), 8 out of 57 (14%), 5 out of 60 (8%), and 7 out of 59 (12%) in males in the control group and at the lowest, intermediate and highest dose, respectively. The historical-control range for this tumour at the testing laboratory was 1.8–8.5%, but a partial review of studies reported recently in the literature revealed a prevalence of 0–17% in control males with several values being ≥ 8%. More importantly, the incidences of islet cell adenomas clearly did not follow a dose-related trend in the treated groups of males, as indicated by the lack of statistical significance in the Peto trend test. It should be noted that there was also considerable inter- group variability in the numbers of females with this tumour (5 out of 60, 1 out of 60, 4 out of 60 and 0 out of 59 in the control group and at the lowest, intermediate and highest doses, respectively). There was no evidence of dose-related pancreatic damage or pre-neo- plastic lesions. The only pancreatic islet cell carcinoma found in this study occurred in a male in the control group, thus indicating a lack of treatment-induced neoplastic progression. Taken together, the data support the conclusion that the occurrence of pancreatic islet cell adenomas in male rats was spontaneous in origin and unrelated to administration of glyphosate.’

The authors seem intent on on telling us one side of the story each time. A more accurate claim would be ‘Glyphosate did not significantly increase pancreatic islet-cell adenoma in male rats in two studies. Any appearance of this form of cancer was due to random chance.’

Claim 3. ‘A glyphosate formulation promoted skin tumours in an initiation-promotion study in mice.’

Here researchers investigated to see if contact with glyphosate could lead to tumour development. Largely the study gave negative results (Yay!) but in two groups there were positive indications of tumour development (Boo!) and one of these groups contained glyphosate (Yikes!). The only problem is that both of these groups were going to initiate cancer. They contained the addition of a known carcinogen that is added to the test to indicate what a positive result. Also to see if the chemical will promote cancer development. By itself glyphosate was negative and did not promote skin tumours in the absence of a carcinogen. This claim would be more accurately written as ‘A known carcinogen (dimethylbenz(∂)anthracene) acted no differently in the presence of glyphosate, however glyphosate in isolation had no effect’. Or even better yet! ‘Glyphosate is not an appropriate cancer cure’.

The IARC monograph even had this to say of the study.

‘The glyphosate formulation tested appeared to be a tumour promoter in this study. The design of the study was poor, with short duration of treatment, no solvent controls, small number of animals, and lack of histopathological examination. The Working Group concluded that this was an inadequate study for the evaluation of glyphosate’

The only two groups that made a tumour were the positive control group with a known cariogenic and the said same cariogenic with glyphosate, in essence another positive control as it acted no differently. I mean really? Compare what the short report said to the full monograph. What the hell is this then doing in the de facto press release regarding your findings IARC? You said it was inadequate! To me this report has the grease of a PR department fingers all over it wanting to spice up the findings. So, to sum up on the animal work front, we have a mixture of non-significant findings been reported as increases in cancer rates. A report which ignores the perfectly well founded and logical findings of a previous WHO report and published literature. And finally what I can only surmise as cherry picking of citing evidence which they themselves deemed inadequate.

Now let’s review; is the IARC’s classification of ‘sufficient‘ evidence in animals warranted? There are more studies in the monograph, but time and time again if you go back to the primary literature you find a lack of significant results. I would strongly argue that this is rather damning on the account of the IARC’s classification criteria. By applying similar criteria they could have released a press release stating the MMR was the probably cause of Autism, the evidence is just as strong.

On to Human work of which there are two claims.

Claim 4. ‘There was limited evidence in humans for the carcinogenicity of glyphosate.’

 The main claim is that there is ‘limited‘ evidence that it (Glyphosate) may play a role in the development of NHL. To say this evidence is limited is to do it a great service, it is as barren as Hull (sorry Hull). The lancet press release cited 3 studies and makes mention of a 4th cohort study (why no citation for evidence that goes against your conclusions IARC?). The 3 cited studies are; Mcduffie et al (7), Ericsson et al (8), and De Roos et al (9). Of all 3 the lancet report begins with this.

‘Case-control studies of occupational exposure in the USA (De Roos), Canada (Mcduffie), and Sweden (Eriksson) reported increased risks for non-Hodgkin lymphoma that persisted after adjustment for other pesticides.’

Let’s start with Mcduffie et al (7) because I like the name, that and the 2006 WHO report had this to say of it.

‘Widely used pesticides, like glyphosate, have recently become a focus of epidemiological research. In the past few years several epidemiological studies have been published that reported weak associations of glyphosate with lymphopoeitic cancers.* However, the results of these studies do not meet generally accepted criteria from the epidemiology literature for determining causal relationships. Generally, the associations were rather weak and rarely statistically significant. Control for potential confounding factors, including other pesticides, was not possible owing to limited available information and small numbers of subjects. It was not measured whether there actually was any internal exposure or the extent of such exposure and, accordingly, a possible dose–response relationship could not be evaluated.’

(N.B * I’ve taken out the list of citations here for sake of brevity)

Huh… No adjustment for other pesticides, that is not what the lancet press release said. Ok, how about old Eriksson et al (8). This study came out after the 2006 report so this time I can not fall back on the WHO’s own 2006 report. What did they do? The number of cases studied for Glyphosate was 29 exposures and 18 controls. 12 people were in the less than 10 days exposure and 17 in more than 10 days. I could point out the limited power in this study but you could then accuse me of bias so instead here is the summary from the monograph.

‘This study had limited power to detect an effect, and there was no adjustment for other exposures’

We now have two citation that did not adjust for other exposures even though the claim that they did in the short press release. Further they mention that the study had limited power, which it does.

Finally, another cool name, De Roos et al. (9) The only study that attempted to correct for other exposures. By the way none of this work is badly done or inadequate, instead I feel it is being forced to do a job it was never designed to do. De Roos et al (2003) pooled results from a number of other trials across the US. For Glyphosate they had 36 exposed individuals and 61 controls. Only 3.1% of the total control group (1933) were exposed to Glyphosate. The IARC group said that they felt this was a large population-based study. However to me it suffers from the same problems of the previous two citations. It is woefully small, the vastness of this study was just in looking at lots of different pesticides. The actual research on Glyphosate is limited to 36 exposed individuals from 3 separate studies. If only we had a large cohort study… oh wait we do! The Agricultural Health Study (AHS) has been examined and re-examined by several different research groups. As the authors say in the press release.

‘The AHS cohort did not show a significantly increased risk of non-Hodgkin lymphoma’.

On the scale of power, cohort studies are better than case control studies (like the 3 studies mentioned here). So we have a large cohort study, in the De Roos analysis of 2005 (10), which is not the same citation as his other work, there are over 15000 exposed cases v.s 13000+ non exposed cases v.s a small case control study (36 v.s 61). It is even the same author so you can’t accuse me of picking sides! No sign of NHLs in the analysis, only multiple myeloma has an increased risk that is due to incomplete data as stated in the monograph.

I should also mention work done, in part, by Monsanto. Mink et al published Epidemiologic studies of glyphosate and cancer: A review, in 2012. This, as it suggests, reviewed a number of publicly accessible studies. They only included results which are publicly available, so if you wanted you could do the exact same study. They found..? No link, surprisingly! I do think scientists are able to remove their bias and I would find it hard to believe that someone anti-gmo hasn’t gone through their analysis with a fine tooth-comb. Interestingly the IARC did not feel the same and left this review out of the monograph while including other reviews in the animal work section. They even went as far to pick out studies which this review had included that are not accessible to the public and disregard them. Something which Mink et al had also done. Maybe the authors heard that Minks lead to a positive trend for snow once and decided it was too risky.

The last Section is on exposure, the press release has a single citation for this entire paragraph. A 2009 report which is summed up as

Claim 5. ‘One study reported increases in blood markers of chromosomal damage (micronuclei) in residents of several communities after spraying of glyphosate formulations.’

There is a slight problem with this, one of the authors of the paper had this to say of the lancet press release.

“When we looked at the differences in the micronuclei between those two groups, we found no difference.

They (IARC) got this totally wrong. They said the study showed there was a relationship…. It’s certainly a different conclusion than the one we came to.” (11)

So yet again the IARC has its trousers down and is implying relationship which have not been shown to exist. Each and every claim in the press release that has a citation can be dismissed as the authors of the report, deliberately or not, miss-representing the science and in doing so, tarnishing the reputation of the WHO. Once again, and I hope after reading through this you agree, this level of research has no right to be called scientific and is on par with charlatans and quacks.

Now this all said and done I’m not at any point going to say Glyphosate is as safe as say water! It is a herbicide, it has chemical properties that make it useful for reducing the parasite load on crops (or what ever farmers call it). You should wash it off your skin if you get it on you (although the absorption though skin looks quite low to me but I’m not an expert). BUT! To say there is sufficient evidence of cancer in animal experimentation and some in humans is just plain silly. Don’t drink Glyphosate, there is good evidence you will get diarrhoea and I’m sure it can be toxic. I do not know if glyphosate will give you cancer. All I’m happy to state is having reviewed the evidence used I do not agree that it ‘is probably carcinogenic’. If this is the level of evidence the IARC and by extension the WHO accepts I have a lot more questions regarding the validity of their other work. I have serious questions for the lead authors (Aaron Blair chair of the monograph group and Kathryn Z Guyton for the lancet press release) regarding their level of scrutiny whilst drafting the press release and how this could be published and used by such an organisation.

Update: California’s EPA have just listed an intent to label roundup as a carcinogen, if this is in part based on the IARC report they should reconsider. Though  roundup and glyphosate are not the exact same thing, glyphosate is an active compound in roundup.

References. I’ve only included open access references, the lancet article requires you sign up but is free.

  1. Lancet press release
  2.  2006 WHO report
  3.  Full Monograph here
  4.  1985 EPA report part a
  5.  Part b
  6.  1986 EPA report
  7.  Mcduffie et al
  8.  Eriksson et al
  9.  De Roos et al 2003
  10.  De Roos et al 2005
  11. Toxicologist statement regarding his paper and how the IARC misrepresented it

Posted In:


  1. Thanks for writing this. I farm in Canada and have been frustrated by how this story has played out in the press. This summer, the Canadian Broadcasting Corporation (CBC) ran a story about a young mother who lives in a house backing on to a farmer’s field. She was sunbathing on her back deck with her baby when the farmer drove by spraying the field. Although none of spray reached her, the story described how terrified she was by the possible exposure of her infant to dangerous chemicals. The reporter did note that the farmer had observed all legal critical distance requirements, but then observed that, since the WHO had recently declared glyphosate a probable carcinogen, perhaps these regulations needed to be changed. No mention, of course, that by sunbathing, the mother was exposing her child to solar radiation, a known carcinogen, and no discussion of dosage, exposure rates or any of the other critical factors involved in chemical safety issues. It just came down to “the WHO said this causes cancer – you should be scared – we need to ban this”.

  2. Very interesting write-up. Thank you for spending time on this issue.

    This might be a minor thing, but why not write “Definitely Not!” instead of “Probably not!” in the title? Would you have felt comfortable writing that?

  3. I’m wondering if you could add a bit about the selection criteria of the articles? in the podcast you seemed to go on for some time about the limitations on what articles the WHO can accept in their reports and I found this incredibly interesting and wondered whether you had a written version explaining it?

      1. Looks like I can not do that because @theleagueofnerd does not follow me. Follow @the_ttguy to fix that – or apparently there is setting that allows you to accept DM from non-followers.

  4. I have skimmed the “monograph” now myself. It has a heap of tables with risk assessments/ odds ratios and 95% Confidence intervals on these. Is it fair to say if the 95% CI for the odds ratio spans 1 then the data is not statistically significant? If that is the case then out of the at least 20 studies they review there are only 3 or 4 that have odds ratios that do not span 1. And then they kinda focus on those to decide that glyphosate “probably causes cancer”. You could equally look at the rest of the data and say that it probably does not cause cancer. It would be useful to know what an odds ratio and CI for well known carcinogens looks like. eg I wonder what the OR is for smoking?

  5. I think that your text on “claim 2” is not correct:

    Claim 2. ‘Glyphosate increased pancreatic islet-cell adenoma in male rats in two studies.’
    Here we have a similar story to claim 1, The levels of adenoma was within normal range for those mice nor were the results significant and the final nail: they did not progress into cancer. The IARC monograph had said this of the studies.
    ‘The range for historical controls for pancreatic islet cell adenoma reported in males at this laboratory was 1.8–8.5%. The Working Group noted that there was no statistically significant positive trend in the incidence of these tumours, and no apparent progression to carcinoma.’

    Pancreatic adenomas rates in male rats in unpublished Monsanto studies from the 1980s are shown in Table 1 of this document:

    You will note that the non-control rats showed rates of 18%, 10%, and 15% and that the p-value on the first is 0.018, which is said to be significant in science, where anything < 0.05 is generally significant. Moreover, if they used a trend fit method that expected monotonicity, then the other p values would be higher if they used a different expectation like a threshold response, which it probably was anyway.

    So, the League of Nerds got this one wrong. Factually wrong, i think…. at least from the numbers in the EPA document. We could be looking at different things, so i'm willing to stand corrected but would like to know your response to the data in Table 1 of the 1991 EPA document, anyway. Thanks.

    Second thing — the adenomas in this case did not progress into cancer because… the rats were sacrificed and autopsied for the study.

  6. League of Nerds is 100% full of shit. I had a long dialog — if you can call the slimy interaction that — on YouTube with Jim Evo in which he slithered away from the evidence in the 1991 EPA memo. He’s nothing but a Monsanto apologist and rationalizer. He’s the kind of a guy who would rationalize the Holocaust during that time, it seems, judging from his ability to doublethink. Anyway, don’t trust these lapdogs of the industry. They lack integrity when it comes down to the details. He’s the guy who can explain away tumors in rats and say “Oh it’s nothing because footnote 17 says that it’s just a little bit more than the maximum historical levels… no nevermind the control group, that doesn’t mean nuthing, just look over here… no don’t look at the little detail that says “cancer” it’s not important…” master propagandists… need to get their game blown apart.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s