Accounting for Multiple Comparisons in Statistical Analysis of the Extensive Bioassay Data on Glyphosate

Area of Science:

• Toxicology
• Carcinogenesis
• Environmental Health

Background:

• Glyphosate is a globally prevalent herbicide.
• The International Agency for Research on Cancer (IARC) classified glyphosate as a probable human carcinogen in 2015.
• Concerns exist regarding the potential carcinogenicity of glyphosate in experimental animals.

Purpose of the Study:

• To re-evaluate the carcinogenicity of glyphosate in rodent bioassays.
• To address statistical limitations in previous analyses of glyphosate cancer data.
• To provide a more accurate assessment of glyphosate's animal carcinogenicity.

Main Methods:

• Analysis of 10 glyphosate rodent bioassays using a multiresponse permutation procedure.
• Adjustment for a large number of tumor types to control false-positive probabilities.
• Evaluation of permutation tests based on p-values from dose-response trend tests.

Main Results:

• False-positive probabilities for carcinogenicity were calculated (e.g., .06, .08 for p ≤ .01).
• More evidence for negative dose-response trends was observed than positive trends.
• No strong statistical evidence for glyphosate being an animal carcinogen was found.

Conclusions:

• The discrepancy with IARC findings is attributed to IARC's failure to account for multiple testing.
• This study provides a more robust analysis of glyphosate animal carcinogenicity data.
• Glyphosate does not show strong evidence of being an animal carcinogen based on this re-analysis.