A Keyword-Based Approach for the Identification and Classification of Mechanistic Evidence for Placental Toxicity Following Exposure to Polychlorinated Biphenyls
On this page:
Mechanistic evidence has the potential to inform biological events linking chemical exposures to developmental health outcomes, such as fetal growth restriction, miscarriage, and stillbirth. Polychlorinated biphenyls (PCBs) are a chemical class of 209 individual synthetic compounds used in numerous industrial applications in excess of 600 million kgs from the 1930s – 1970s in the U.S. Although PCBs were banned in the U.S. in 1979, humans are still exposed via inhalation of volatilized PCBs or contaminated dust, contact with contaminated dust or with primary or secondary sources of PCBs, and ingestion of foods contaminated with PCBs, including breast milk. Thus, adverse developmental health effects following exposure to PCBs during sensitive periods of development are of high concern. An extensive body of literature describes the associations between PCB exposure and decreased offspring weight/growth and increased offspring mortality, However, mechanistic information to support the biological plausibility of these effects have not been systematically reviewed. One of the challenges in identifying and classifying the mechanistic evidence is the sheer size of the PCBs literature base (approx. 70,000 studies generated by EPA). Given the integral role of placental function in both offspring growth and survival, a targeted, keyword-based classification approach was developed to expedite the identification, analysis, and prioritization of mechanistic literature supporting biological plausibility of adverse developmental outcomes resulting from placental toxicity following exposure to PCBs. A list of 67 keyword entities relevant to placental health (e.g., trophoblasts, preeclampsia, JEG-3, labyrinth) was developed by subject matter experts and further refined using controlled ontologies (e.g., Unified Medical Language System). The keyword entities were searched against EPA’s PCB literature database, which resulted in the identification of 859 studies. Title and abstract screening was conducted in DistillerSR software and relevant studies were categorized based on nine types of mechanistic evidence (e.g., disrupted placental hemostasis, disrupted placental angiogenesis, oxidative stress in placental cells/tissues, altered inflammation/immune response in placental cells/tissues). These tags were determined a priori based on subject matter expert knowledge of mechanisms of placental toxicity and a tenth tag of “Other (describe)” was included to ensure mechanisms falling outside the predefined tags were captured. This process identified 48 studies, with endocrine disruption, altered trophoblast fate, and altered angiogenesis being the most frequently reported mechanisms. These 48 studies will be further reviewed by subject matter experts, and relevant mechanistic evidence will be synthesized to assess the potential for placental toxicity to mediate adverse effects of PCB exposure on offspring growth and mortality. We estimated the automated keyword approach saved 2,317 hours of time to review the literature database compared to the standard approach of implementing electronic methods to actively prioritize studies during initial screening, resulting in a 96% reduction in time required to complete the review of literature relevant to placental effects associated with exposures to PCBs.