Application of the Navigation Guide systematic review methodology to the evidence for developmental and reproductive toxicity of triclosan

Abstract

Background: There are reports of developmental and reproductive health effects associated with the widely used biocide triclosan.

Objective: Apply the Navigation Guide systematic review methodology to answer the question: Does exposure to triclosan have adverse effects on human development or reproduction?

Methods: We applied the first 3 steps of the Navigation Guide methodology: 1) Specify a study question, 2) Select the evidence, and 3) Rate quality and strength of the evidence. We developed a protocol, conducted a comprehensive search of the literature, and identified relevant studies using pre-specified criteria. We assessed the number and type of all relevant studies. We evaluated each included study for risk of bias and rated the quality and strength of the evidence for the selected outcomes. We conducted a meta-analysis on a subset of suitable data.

Results: We found 4282 potentially relevant records, and 81 records met our inclusion criteria. Of the more than 100 endpoints identified by our search, we focused our evaluation on hormone concentration outcomes, which had the largest human and non-human mammalian data set. Three human studies and 8 studies conducted in rats reported thyroxine levels as outcomes. The rat data were amenable to meta-analysis. Because only one of the human thyroxine studies quantified exposure, we did not conduct a meta-analysis of the human data. Through meta-analysis of the data for rats, we estimated for prenatal exposure a 0.09% (95% CI: -0.20, 0.02) reduction in thyroxine concentration per mg triclosan/kg-bw in fetal and young rats compared to control. For postnatal exposure we estimated a 0.31% (95% CI: -0.38, -0.23) reduction in thyroxine per mg triclosan/kg-bw, also compared to control. Overall, we found low to moderate risk of bias across the human studies and moderate to high risk of bias across the non-human studies, and assigned a "moderate/low" quality rating to the body of evidence for human thyroid hormone alterations and a "moderate" quality rating to the body of evidence for non-human thyroid hormone alterations.

Conclusion: Based on this application of the Navigation Guide systematic review methodology, we concluded that there was "sufficient" non-human evidence and "inadequate" human evidence of an association between triclosan exposure and thyroxine concentrations, and consequently, triclosan is "possibly toxic" to reproductive and developmental health. Thyroid hormone disruption is an upstream indicator of developmental toxicity. Additional endpoints may be identified as being of equal or greater concern as other data are developed or evaluated.

Keywords: Antibacterial; Personal care; Policy; Risk assessment; Soap; Thyroid.

Figure 1

Search results

Figure 2

Risk of bias results for included human studies. The risk of bias results did not differ according to different specific outcome measures within the studies, and therefore the results are presented by study. Cullinan et al. was a subset of a randomized controlled trial and was therefore evaluated under the “Randomization” risk of bias domain, and received a “N/A” (not applicable) rating under the “Baseline differences” domain. The other studies received a “N/A” rating under the “Randomization” domain.

Figure 3

Risk of bias results for included non-human studies. Note Stoker et al. presents two experiments using two separate cohorts (pubertal assay and uterotrophic assay); each cohort was evaluated for risk of bias separately. *Stoker et al. pubertal assay cohort assessed free T4, total T4, and TSH. **Stoker et al. uterotrophic assay cohort assessed free T4 and total T4.

Figure 4

Prenatal triclosan and thyroxine

1. Prenatal triclosan administration and thyroxine concentration as a percentage of the control group for doses up to 300mg/kg/day

2. Prenatal beta-estimates for dose response and the random effects meta-analysis estimate

Abbreviations: PND, postnatal day; GD, gestational day; LS, litter size. The vertical gray bar in A represents the line of no effect (the control group normalized to 100%); horizontal error bars represent 95% confidence intervals; in B, symbol sizes represent the log of the weight in the meta-analysis.

Figure 5

Postnatal triclosan and thyroxine

1. Postnatal triclosan administration and thyroxine concentration as a percentage of the control group for doses up to 300mg/kg/day

2. Postnatal beta-estimates for dose response and the random effects meta-analysis estimate

Abbreviations: PND, postnatal day; GD, gestational day. The vertical gray bar in A represents the line of no effect (the control group normalized to 100%); horizontal error bars represent 95% confidence intervals; in B, symbol sizes represent the log of the weight in the meta-analysis.