The report you refer to is by Paz-y-Miño et al. (2007), published in the journal Genetics and Molecular Biology, a small Brazilian journal (impact factor 0.73, so not a well-recognized journal). César Paz-y-Miño has an OK publication record and studies a number of regional issues using his expertise. This report assesses "DNA damage" using what's called a "comet assay," an assay in which cells are placed into an agar matrix and subjected to an electric field. DNA is charged, so it moves to the positive pole. Damaged DNA moves faster — that's the basis of the assay.
In this report, 24 people from an ag-intensive area where glyphosate was used were compared with 21 in an area 80 km away. Blood was drawn "between 2 weeks and 2 months" after glyphosate application to the crops. There is one table of data showing the DNA from those living near the farm (50 percent tested were 200 m–3 km).
The results show consistently higher migration in the "exposed" group, suggesting more damage, according to the authors.
Before we get too excited about the results:
1. Glyphosate moves quickly from the body. After two weeks, there would be a negligible effect, if any, from acute exposure. The samples could have been from people tested two months after exposure; the authors don't specify.
2. The authors say that the "exposed" group had sprays directly over their homes in 50 percent of cases, and that applications were "20 times the maximum recommended application rate for the formulated product, which may explain our comet assay results."
So, directly spray the homes of the workers with 20 times the normal concentration? Hmm.
3. Glyphosate is rapidly removed from the body, and in no case has it been demonstrated to damage DNA or even be carcinogenic (it is classified as "not carcinogenic" by even the strictest standards). What is happening?
Here are additional considerations and interpretations:
First, if these workers were tested 14–60 days after being sprayed with 20x glyphosate, what else are they spraying down there imprecisely and at levels far above the recommended ones? Are these people chronically ill from prolonged exposure to ag chemicals? That could be a better explanation.
But the best explanation is "Blood samples (from the unexposed group) were collected and processed as for the exposed group, but not concomitantly."
Bingo. The authors counted on a single replicate that was processed at different times. How the blood was handled, how it was prepared — all could easily account for the results seen. The fact that it was one replicate is also quite telling. I'd never publish with fewer than three on this kind of test.
The best thing that could be said is that the data show a potential starting point. It would have been good to see the data and have the controls and treatments collected and processed blindly and at the same time.
Conclusion: Maybe good work, maybe not. Maybe trustworthy data that are a hint of things to study further, maybe not. My guess is not. It has been seven years since this study, and no further evidence to support the DNA-damage conclusion has emerged. In a 2011 report by the same authors, glyphosate showed no effect in DNA damage in a larger test with greater resolution in Colombian/Ecuadoran populations.
The most likely explanation of the findings is that the cells in one group broke down or had some other damage during handling that led to the results. That's why there has been no follow-up on this study.