Again, here is the PRIMARY ARTICLE that the articles reference:
http://www.sciencedirect.com/science/article/pii/S0278691512005637
I'd like to point out the Herald Article is a press release written by the "Sustainable Food Trust"- which is an organic foods movement group:
http://www.sustainablefoodtrust.org/
So, let's assume the Press Release is fairly biased and that those that wrote it have modest scientific literacy.
On to the paper!
Compressing the whole thing into a few digestible sentences isn't doing anyone favors, but that's what I'll try to do. I'd encourage people to read the actual article.
I think the idea of growing corn, spraying it, and feeding animals diets consisting of 11, 22, and 33% of that corn is gimmicky. If the hypothesis is: "Round-up is cytotoxic and carcinogenic at currently consumed levels" then you feed them controlled doses of Round-up and an identical balanced diet - which they did in addition to the silly corn experiments, which don't account for the difference in nutrient intake over 2 years.
So anyhow, I think that part of the study is uncontrolled. They also looked at "200 rats", but broken into 10 rats / group, where group is feed-type and sex split. So for any given treatment they only have 10 rats to gain statistics on, which anyone with a relative statistical background can tell you is insufficient for any analysis if your control group is also presenting with effects (untreated rats died and acquired tumors during the study). Also if you look REAL CLOSE you'll realize that they only actually tested 10 rats of each sex for the "no treatment" subgroup. It's the same "0" treatment data on each graph. So in total they looked at 20 rats for the null treatment to compare to 180 rats of various other treatment types. Bummer.
So if we discount that any given dose-set is the sum of 10 animals of the same sex, and want to get anything out of this study, we want to look at the animals fed water laced with Round-Up. That's where the data is useful. So let's look at that.
Group A: Water +
.00000001% Round-Up [amount found in some tap waters]
Group B: Water +
.09% Round-Up [amount found in some US feed]
Group C: Water +
.5% Round-Up [working dilution used to spray crops directly]
I'm not crazy about the idea of feeding animals straight from the crop-duster dilutions for two years to prove a point (group C), but I see where they're going with A and B. I'm not sure that Group C has any real-world relevance, unless some farmer is getting really thirsty out in the field. Also, this brings me to an aside regarding controls. Untreated is great, but positive and negative controls are also informative. I imagine feeding rats Water + 0.5% mineral oil for 2 years would cause oncogenic phenotypes. The best experiment would have been to feed mice known environmental carcinogens or inert substances at the same doses and compared the relative carcinogenic index of Round-Up.
For males there's no real effect. Straight off the plane pesticide for two years caused metastasis in 1-2 rats. Not sure what the spontaneous metastasis rate in these rats is, would need more untreated control mice to know if that's even relevant. Something odd to note. Figure 1 shows 1 rat in Group A needing to be put down due to huge tumor growth, but in Figure 2 none of the Group A mice were documented as acquiring anything apart from small internal tumors. So there's a data disconnect there.
The female rats are weird. Even untreated rats acquired tumors so large they had to be put down before the 2 year period was up. This isn't exactly the "control" group I'd want to use to prove carcinogenicity of a substance. Even so, there's no real difference between trace amounts of Round-Up and 100,000x that amount, some metastasis in the Group C - but again I'm unaware of whether that's a spontaneous occurrence in the normal mice.
The same statistical and analysis problems arise in the organ and tissue studies. Not enough numbers, too few metrics, not enough divergence between extreme dose ranges to presume it's the Round-Up.
- The experiment should be repeated.
- A cohort of at least 25 animals per condition should be tested (200 animals total), though I'd rather cut out condition C and test 50 animals per condition.
- I'd pick a less tumor susceptible background strain.
- The water intake should be controlled such that each mouse takes in the same total amount of treatment per day.
- Positive and negative controls should be used to demonstrate degree of significance of Round-Up and sensitivity of the animals to tumor acquisition.