2.2. Criteria for Study Inclusion

In order to examine consistency of chemical effects in multiple Hershberger studies, we defined criteria for including/excluding studies from this analysis based on their adherence to guideline study protocols. In contrast to the similar undertaking examining increases in uterine weights following test chemical administration in published uterotrophic studies (Kleinstreuer et al. 2016), the Hershberger assay is a considerably more complicated experimental design. The OECD/US EPA Hershberger guideline includes five AST weights, several optional organ weights, and has both androgenic and anti-androgenic portions. The final guideline Hershberger protocol treats castrated peri-pubertal male rats with test chemicals for 10 consecutive days, allows at least seven days of post-surgical recovery prior to dosing, includes six or more rats for each dose level, measures weights of five ASTs (VP, SV, LABC, CG, and GP), and conducts the necropsy 18–36 h after the last dose (Table 1). At least two dose levels are included in the androgenic mode of the assay and at least three dose levels are included in the anti-androgenic mode. The anti-androgenic mode includes coadministration of test chemical with 0.2 to 0.4 mg/kg bw/d TP, and appropriate positive and negative control groups for both the androgenic and anti-androgenic modes are required (Table 1). A significant effect of a chemical tested in the androgenic mode of the assay is an increase in the mean weight of two or more ASTs compared to vehicle controls (i.e. a “positive” androgenic effect). Similarly, a chemical is considered to have a significant effect in the anti-androgenic mode if coadministration of test chemical plus reference androgen results in significantly decreased weights of two or more ASTs relative to reference androgen controls (i.e. a “positive” anti-androgenic effect). Thus, the lowest observed effect level (LOEL) is the chemical dose that causes a significant change in at least two ASTs. The no observed effect level (NOEL) is also noted for the chemical dose at which no effect occurred (or only occurred in one AST) or the highest concentration tested where fewer than two AST weights were affected. The LOEL and NOEL are reported in mg/kg bw/d (Supplemental Table 2).

Methodological descriptors for the Guideline Hershberger assay and criteria considered here for positive in vivo androgenic and anti-androgenic results and negative results with no significant in vivo effect on >1 accessory sex tissue weight.

PND = postnatal day, SC = subcutaneous, IM = intramuscular, IP=intraperitoneal, TP= testosterone proprionate, TC = testosterone cipronate, MT= methyl testosterone, VP = ventral prostate, SV=seminal vesicles, LABC=levator ani bulbocavernosus, CG=coagulating gland, GP= glans penis, AST=accessory sex tissue

Similar to the evaluation of chemicals tested in uterotrophic studies (Kleinstreuer et al. 2017; Browne et al. 2015), we developed a priori criteria for including/excluding Hershberger data from this analysis to determine consistency of responses and identify candidate in vivo reference chemicals. Proposed criteria were reviewed with the Reference Chemical Working Group, convened by the Interagency Coordinating Committee on the Validation of Alternative Methods (ICCVAM; https://ntp.niehs.nih.gov/pubhealth/evalatm/test-method-evaluations/refchem/index.html), specifically to identify candidate reference chemicals for novel toxicological method validation. The Hershberger guideline protocol was optimized for sensitivity following several rounds of validation through the OECD, during which high quality data were generated (OECD 2008b). In the interest of including all available high quality data, the group agreed that results from chemicals tested in Hershberger protocols demonstrated to be reproducible and reliable though OECD test method optimization and validation, but less sensitive than the final guideline method, should be considered in the circumstance where positive results were observed. The underlying logic was that if a test chemical resulted in a positive result (i.e. AST weights were significantly different from control animals) using a less sensitive protocol, then the same chemical would likely have a positive effect when tested in a more sensitive study design or animal model. Candidate negative reference chemicals (i.e. chemicals that did not produce a significant change in at least two AST weights compared to controls) were only considered from study designs that more closely adhered to the final guideline protocol. Criteria for the Hershberger test guideline and those used to identify chemicals with positive and negative results are summarized in Table 1.