Table 2.
Criticisms of the studies of Tyl et al. (2008a, 2008b, 2008c) and responses
| Criticism | Response |
|---|---|
| 1. Myers et al. questioned the use of CD-1 Swiss mice because of their “aberrant insensitivity” to estrogens. | Recently, Ryan et al. (2010a) used a wide range of doses from 0.05 to 50 µg/kg bw/day for the positive control estrogen, EE2. They observed a comparable sensitivity for CD-1 mice, Long-Evans, and Sprague-Dawley rats for several adverse effects. Therefore, the use of CD-1 mice by Tyl et al. (2008a, 2008b, 2008c) is justified. Tyl et al. purchased their CD-1 mice from Charles River to avoid supplier effects. It should be considered that also vom Saal and some of his colleagues used CD-1 mice and reported low dose effects (Timms et al., 2005). Low-dose effects of BPA were also reported in Long-Evans (Akingbemi et al., 2004) and in SD rats. |
| 2. Myers et al. considered the prostate weights reported in the study of Tyl et al. (2008a, 2008b, 2008c) “abnormally high” and suggested “that the dissection procedures for the prostate in the Tyl laboratory included other nonprostatic tissues in the weight measurements, rendering them unusable.…” This criticism was also expressed by Gies et al. (2009). | Tyl (2009a, 2009b) reported that paraffin blocks and slides of the analyzed prostates are still available and indicate no evidence of extraneous tissue/fat or excessive inflammation (all recently audited by the US FDA). In a response letter (Tyl, 2009a, 2009b), the author documented that considering the age of the mice, prostate weights in their study were within the range of published data. Differences from vom Saal's study may be explained by his post wean caging regimen. Vom Saal's CF-1 mice were housed by sex. Pups were weaned at 23 days of age and male littermates were housed three per cage (Nagel et al., 1997). Prostate weight was determined when the mice were 6 months old. For this purpose, one male per litter was randomly selected and individually housed for 1 month before determination of prostate weights (Nagel et al., 1997). This procedure is problematic because the dominant cage male develops large androgen-dependent accessory organs, whereas subservient cage males have smaller prostates. It is questionable whether 1 month of single housing may compensate for months of group housing and its influence on sexual development of male mice. In the study of vom Saal (Nagel et al., 1997), no controls were presented with respect to the influence of group housing and compensation by single housing and the authors did not report whether they determined prostate weight only from the dominant males or if all animals were included. Since the authors wrote that “one male per litter was randomly selected…” (Nagel et al., 1997), they likely included both, dominant and subservient males. This may explain the high variability of the prostate weights in studies from vom Saal's lab compared to much lower variability in Tyl's studies (Tables 1 and 2 in Tyl, 2009a, 2009b). It should also be taken into account that the number of animals exposed to BPA in vom Saal's study (n = 7) is very low also, considering that an additional confounder (dominant versus subservient mice) may be present. Much higher numbers of animals we re used in Tyl's studies (28 per group). After careful evaluation of Tyl's studies, the criticism of Myers et al. (2009) and Gies et al. (2009) concerning prostate weights is unsubstantiated. However, several aspects may be critical in vom Saal's studies, such as the influence of dominant versus subservient males, no available histology, and use of a now obsolete in-house strain preventing other researches to reproduce the initial positive studies. Another drawback is the statistical analysis.Concerning the criticism that crude mistakes were made by Tyl's laboratory in determining prostate weights, it should also be considered that this laboratory has carried out a large number of reprotox studies (often sponored by the US Government) and has successfully joined interlaboratory validation studies, where the Tyl laboratory achieved some of the most precise prostate weight data and robust treatment-related effects (EPA review). |
| 3. Myers et al. criticized the high incidence and severity of prostatitis in the animals from Tyl's study, which may compromise their results. | The paraffin slides of the prostates of the mice from Tyl's study are still available and have been reanalyzed by an independent pathologist. The results have been published (Tyl, 2009a, 2009b). The incidence of inflammation in CD-1 mice of Tyl's study matches the low incidences of prostatitis seen in many mouse strains without treatment. Therefore, it is extremely unlikely that the results of Tyl et al. (2008a, 2008b, 2008c) have been compromised by prostatitis. It should be kept in mind that vom Saal's laboratory did not include any histopathological examinations (Nagel et al., 1997). Therefore, a possible confounding influence of prostatitis could not be accounted for in their study. |
| 4. Myers et al. criticized that the diet used in Tyl's study contained phytoestrogens, which they claim would interfere with BPA activity. | Recently, Tyl et al. have published the genistein, daidzein, and glycitein contents of the standard diet (Purina Certified Ground Rodent Chow No. 5002) used in their study. Vom Saal and colleagues have not reported the phytoestrogen content of their diet. The majority of “normal rodent diets have similar levels of phytoestrogens.” Although it is not possible to compare diets between Tyl's and vom Saal's laboratories, it is extremely unlikely that the diet in Tyl's study compromised an estrogenic response, because in studies with estradiol in which mice were fed the same standard diet as those in the BPA studies, estradiol (0.5 ppm) clearly accelerated acquisition of puberty (Tyl, 2008a). |
| 5. Myers et al. criticized that “Tyl et al. (2008a) did not examine any neurobehavioral end points,” which Myers et al. interpreted as a “glaring omission of Tyl.” | The multigeneration study performed by Tyl et al. (2008a) represents a standard test performed according to a specific guideline that is not designed to investigate neurobehavioral endpoints. Therefore, there is certainly no “glaring omission of Tyl,” because behavior simply is not an endpoint in this study type. However, there are other studies that have focussed on neurobehavioral endpoints: Ema et al. (2001) observed no effects in their two-generation, low-dose (0.2–200 μg BPA/kg bw/day) study. Similarly, Ryan et al. (2010a) obtained negative results in a detailed study on reproductive development, function, and behavior in female rats exposed perinatally to a wide range of BPA doses (see also Sharpe, 2010). Stump et al. (2010) published the most recent study conducted according to GLP. The results show that BPA, at levels of exposure 4000-fold higher than the maximum human exposure in the general population, does not cause any discernible adverse effects in female rats. By contrast, EE2, used as a positive control in this study, caused major adverse health effects at doses in the range of those applied in early contraceptives (Ryan et al., 2010a; highlight report: Sharpe, 2010). |
| 6. Myers et al. criticized Tyl et al. (2008a, 2008b, 2008c) for using too many animals: “…all of the studies by Tyl et al. were significantly overpowered and this is in direct violation of federal guidelines for conducting animal research, a fact about which U.S. FDA regulators seem unaware.” | The consideration of animal protection aspects should certainly be an important point in all animal testing. But even if we assume that more animals than required were used, according to statistical power analysis, this would not lead to different or incorrect conclusions from a study. However, the numbers of animals used in Tyl's studies are appropriate—28 mice/sex/group/generation. This is in line with the OECD and US EPA toxicity testing guidelines, which require at least 20 pregnant females per group. |