Comment on “Bisphenol A (BPA) in U.S. Food” - Environmental

USFDA, Center for Food Safety and Applied Nutrition, College Park, Maryland 20740, United States. Environ. Sci. Technol. , 0, (),. DOI: 10.1021/es2002...
2 downloads 0 Views 643KB Size
CORRESPONDENCE/REBUTTAL pubs.acs.org/est

Comment on “Bisphenol A (BPA) in U.S. Food” recent article1 caught our attention when the abstract stated “We know of no studies reporting BPA in U.S. fresh food, canned food, and food in plastic packaging in peer reviewed journals.” We are concerned that this and several related statements might misdirect readers about BPA in canned foods. Particularly, that the BPA concentrations reported were biased low and the stated association of BPA concentration and pH is unsupported by the reported data or previous literature. Finally, can information, sample handling procedures, and sufficient method validation and performance data, which might explain the anomalous results, were not reported. We identified 16 peer reviewed publications that measured BPA in food samples from North America. If we discard measurements from Canada, Mexico, and beverages, we still count 11 relevant articles. A closer review was warranted, in part because several statements made in Schecter et al. were directly contradicted by the referenced and unreferenced publications. The BPA concentrations reported in Schecter et al. were consistently lower, on average 1.5 orders of magnitude lower, than the referenced literature. Of the 12 quantitative comparisons Schecter et al. made to literature BPA concentrations, the ranges overlapped for only green beans and did not overlap for the 11 other comparisons. The BPA concentrations reported by the authors for nonoverlapping comparisons were on average 7.1 860 fold lower than previous reports. Comparing BPA concentrations with published data not referenced2 8 also indicates that the concentrations reported by Schecter et al. were low. BPA concentrations for the same condensed soup product were 6 fold lower, whereas ready to eat soups, corn, and tomato paste were an order of magnitude lower than a Canadian survey.3,7 Green beans were the only product where reported concentrations were greater.3,7 On a product-toproduct basis, two infant formulas reported by Schecter et al. were 2 8 and 20 times lower than previous reports.2,5,6,8 Only limited validation and method performance data were reported for the analytical method described. Replicate analyses were performed, but precision data for BPA concentrations were not reported, making it impossible to determine if sample-tosample differences were significant or within the error of the method. Despite the fact that samples were freeze-dried prior to the addition of the internal standard and a large variety of matrices were analyzed, method accuracy and/or spike recovery were not evaluated or reported. Sample homogenization and separation of solid edible portions from liquid brine was not described. However, studies have shown that BPA partitions between the two portions,9 strongly affecting reported concentrations. Without these method performance data it is unclear if the low BPA concentrations reported here are due to variability between products or bias of the method. We found no evidence in Schecter et al. that BPA concentrations varied with pH. We found no linear or quadratic fits, or correlations of pH and BPA (p g 0.36), regardless of transformation, detection limit substitutions, or data subset used. Additionally, there were no differences between foods of different pH (p g 0.1). While the article reported a single difference

A

This article not subject to U.S. Copyright. Published 2011 by the American Chemical Society

between the two most populous pH groups, this was only due to inappropriate detection limit substitution (at 44% nondetect frequency) and inflation of statistical power through the use of triplicate sample analyses. Without differences in BPA concentration between multiple pH groups, or at least a significant correlation, we see no evidence for a relationship between the two variables. Furthermore, the study did not acknowledge or control for packaging characteristics known to affect BPA concentrations. Can size, type (2 or 3 piece), closure type and food type generally correspond with coating type, thickness, and residual BPA in the coatings, all strongly affecting BPA concentrations (2, 8, 10, 11). For example, pineapple cans are usually coated with tin, therefore the reported result (