Over the past several decades, the concept of chemical persistence assessment using multimedia models has been broached many times, but with limited uptake in regulatory settings (Boethling, 2016). However, a unit-world concept remains a useful mechanism to evaluate various types of data that inform fate and degradation processes (Figure 1). Therefore, an integrated persistence-assessment framework is needed based on a unit-world concept to organize and define data relevance in the WoE approach recommended by OECD (OECD, 2019).

Principles of WoE

WoE refers to a decision-making approach in which one collects all available LoEs, triages them by reliability and relevance, and integrates this information into an overall picture to inform risk-based decision making. In the past, this process was viewed as a subjective approach in the face of conflicting or incomplete information but is meant to allow flexibility in contrast to bright-line criteria (Weed, 2005).

The concept of gathering existing information, assessing its reliability in a transparent and systematic manner, and interpreting results in a WoE framework has been discussed extensively over the past decade (Rhomberg et al., 2013; SCENIHR, 2012) including the development of templates to organize data for regulatory activities (link). Lutter et al. (2015) were one of the first groups to propose an approach on how to evaluate WoE frameworks. Their recommendations came from a wider discussion that took place in 2012 among a group of scientific and regulatory experts who were involved with agriculture and pest management. Their proposal included formulating a hypothesis related to specific endpoints under evaluation, providing more transparent discussions of WoE and a more rigorous and systematic determinant of WoE for the evaluation of substances (i.e., consistency). They proposed that this approach should not only include all available LoEs but also an evaluation of data quality and study reliability.

More recently, ECETOC proposed a WoE-based framework applied to the PBT/vPvB (persistent, bioaccumulative, toxic/very persistent, very bioaccumulative) assessment (ECETOC, 2014). Their approach followed the general WoE framework outlined by the Scientific Committee on Emerging and Newly Identified Health Risks (SCENIHR, 2012). Again, building on previous frameworks, the approaches outlined by SCENIHR not only recommended identifying, collecting, and evaluating all sources of information related to the specific question (hypothesis) at hand but also introduced the consideration of scientific quality and relevance of the studies as a critical component of the assessment. This group also recommended the use of a scoring system for the overall WoE assessment (i.e., strong, moderate, weak, uncertain).

In 2013, a group of experts surveyed more than 50 existing WoE frameworks to identify best practices (Rhomberg et al., 2013). They concluded that a WoE approach consists of several critical steps, including evaluating strength and quality of existing data and information, and a systematic process on how to integrate these data to best inform risk. Since then, many different WoE frameworks have been proposed with different phases in this stepwise process but, essentially, most of them adhere to the following logic: (i) problem formulation including hypothesis definition, (ii) data selection, (iii) evaluation of data and study quality evaluation (criteria for review of individual studies), and (iv) data integration.

In a more recent publication, the OECD outlined the key elements necessary for implementing a WoE approach to the evaluation of chemical substances (OECD, 2019), which is used as the basis for the proposed persistence assessment framework in this paper (Figure 1). The OECD (2019) document was being put forward as “good WoE practices” and still accounts for professional judgment as a critical component in constructing the appropriate framework for differing specific endpoints. The OECD provides a stepwise approach to conducting WoE for chemical evaluation including five key elements (Figure 1): (1) problem formulation (hypothesis development), (2) evidence collection, (3) evidence evaluation, (4) evidence weighing, (5) evidence integration and reporting (OECD, 2019). This framework illustrates that developing criteria for study quality (e.g., reliability) and relevance is especially important when multiple values for the same metric are available for a given substance. The study designs and data interpretation should be examined closely to determine the reliability and relevance of the findings as well as sources of variability. WoE is a tool used to evaluate data, but specific applications need specific considerations to appropriately organize and compare different data types. For example, the specific WoE approach taken for persistence assessments will differ from the specific approach needed for toxicity assessments.

Examples of previous WoE approaches to persistence assessment

Before the publication of the OECD WoE guiding principles (OECD, 2019), many of the key elements outlined by the OECD were already applied in chemical safety assessments (Brandt et al., 2016; Bridges & Solomon, 2016; Giesy et al., 2014; Hughes et al., 2020; Wassenaar & Verbruggen, 2021). Giesy et al. (2014) used a WoE approach to assess the organophosphorus pesticide chlorpyrifos. There was a significant amount of data available for this compound, with multiple laboratory test values for half-lives spanning 2 orders of magnitude in various media, partially exceeding persistence thresholds. They used the geometric mean of the half-lives to reconcile the data variation, with the results that the persistence criteria were not exceeded in any media. In addition, they looked at field data, which revealed relatively short half-lives not exceeding persistence criteria. Additionally, chlorpyrifos was known to hydrolyze, albeit with significant variation in half-life at different pH values. The conclusion was that the criteria for persistence were not exceeded.

Brandt et al. (2016) proposed a WoE approach to assess the persistence of a group of substituted phenolic benzotriazoles. They hypothesized that these substances would be very persistent in the environment as defined by their REACH (Registration, Evaluation, Authorisation and Restriction of Chemicals; Regulation [EC] No 1907/2006) endpoints for persistence. There was insufficient experimental information available to reach a conclusion on the persistence of these substances. They used QSARs and biodegradation models for phenolic benzotriazoles to group them as a class and, because of their hydrophobic nature, identified sediments as the most relevant compartment for persistence assessment. Here, they relied on several LoEs, including environmental monitoring of phenolic benzotriazoles in different environmental compartments and biota (literature values), and one laboratory biodegradation and one field dissipation study. The final element of the WoE approach was a sediment core analysis to determine the in situ half-life. Although the authors considered the relevance of the data based on physicochemical properties (and modeling degradation kinetics and pathways), the authors subsequently diverged from the OECD process for evidence evaluation. A key principle in evaluating evidence is to determine data reliability, uncertainty, and relevance. The authors opined that weighting the evidence could not be done objectively and argued for more of a “summary narrative” approach. Here is where the key elements of the OECD guidance may prove helpful in future, because it separates the weighting process from the process of setting more objective quality criteria to assess data reliability.

The OECD document points out that weighing is used to differentiate data sources, not to judge overall quality, which they consider as a separate (earlier) step. Weight can be qualitative or assigned a score and is based on reliability and relevance (e.g., similar to Klimisch scoring; Klimisch et al., 1997). Here, the OECD document states that weighing of data should be based on “clear and transparent” methodology. Suter et al. (2020) provided similar recommendations in their comparison of a WoE versus a systematic review for evaluating multiple LoEs to help inform decision making. They also pointed out that, regardless of the methodology selected, the important component is to follow scientific standards of transparency concerning the choice and implementation of the approach. In the end, Brandt et al. (2016) integrated the multiple LoEs into this WoE approach and concluded that the phenolic benzotriazoles under consideration are persistent in the environment. However, they also suggested that more guidance is needed to provide a more quantitative WoE approach while still maintaining some flexibility for its application.

In Wassenaar and Verbruggen (2021), a WoE approach, as indicated by Chapter R.11 of the ECHA Guidance on Information Requirements and Chemical Safety Assessment, was used to evaluate the persistence, bioaccumulation, and toxicity of alkylated 3-ring PAHs (polyaromatic hydrocarbons). Eighteen studies with persistence data were screened for relevance to determine environmental degradation half-life values. As described in the paper, none of the 18 studies qualified, and therefore there was no WoE for a half-life determination. Instead, a WoE argument was made that the data as a whole indicated a trend of increasing half-life with increasing alkylation for alkylated 3-ring PAHs.

Hughes et al. (2020) also used the same R.11 guidance process to assess the persistence of phenanthrene. The studies were assessed and qualitatively weighted by relevance (similarity to standard test guidelines). In silico half-life predictions, nonrelevant guideline studies, and monitoring studies were also incorporated into the WoE. The consistency or coherence of the data was assessed by assembling all half-lives per compartment and understanding the outliers. Their conclusion was that phenanthrene is not a persistent molecule in any environmental compartment, and they reinforce this using a Pov calculation. The aforementioned environmental half-life data were used as the basis of the Pov example given in the Supporting Information.

A more recent quantitative analysis for the cyclic volatile methyl siloxanes (cVMS; Bridges & Solomon, 2016) most closely follows the key elements outlined in the OECD WoE document (OECD, 2019). Their underlying hypothesis was that the persistence of these substances (octamethylcyclotetrasiloxane [D4], decamethylcyclopentasiloxane [D5], and dodecamethylcyclohexasiloxane [D6]) exceeded the threshold set by the Stockholm Convention. In developing their framework, they relied heavily on previously published WoE frameworks (SCENIHR, 2012). Their analysis consisted of LoE, which included general trends in environmental monitoring measurement, laboratory and field studies, and multimedia fate models where the unique physicochemical properties of the cVMS (low solubility and high vapor pressure) directly affected the overall environmental fate (partitioning) of these materials. Although the half-life in air for the cVMS (abiotic processes dominating) exceeded the two-dimensional criterion for persistence Matthies et al. (2016), reasoned that the tendency of these materials to remain in the atmosphere, where they are degraded, limits their ability for long-range transport and the ability to affect organisms. The authors conclude that these substances are not persistent in a way that would be harmful. The authors did question the appropriateness of biodegradation test setups that prevent evaporative losses, when those are artificial constructs that would not exist in the natural environment.

A key aspect that differentiated the Bridges and Solomon (2016) evaluation, using WoE for persistence, from others was the development of an extensive, transparent scoring criteria for quality, reliability, and relevance. The authors emphasized the importance of assessing the LoE for relevance (to the stated hypotheses) and reliability and that reliability scoring be conducted in an objective, reproducible, and transparent manner. Also, the studies needed to have appropriate quality control or quality assurance, data processing, and clarity to be considered reliable. In the scoring of the field studies, they observed that reliable sampling and good analytical practice were important and identified more than 10 predefined criteria for quality and relevance to laboratory studies. Some of the more relevant quality criteria suggested by Bridges and Solomon (2016) and others (Forney et al., 2001; Hermsen et al., 2018; Koelmans et al., 2019; Kowalczyk et al., 2015; Martin et al., 2017; Moermond et al., 2016; Ott et al., 2020) are outlined in Table 2.

In the end, the results from the Bridges and Solomon (2016) assessment were visualized in graphical plots of score for quality against the score for relevance to each study (Van Der Kraak et al., 2014). This visualization facilitated coherence or integration of all relevant, reliable evidence of P assessment, and the information provided a more holistic approach to the integration of physicochemical properties, multimedia fate modeling, testing, and field observation to assess the overall Pov of cVMSs in the environment.

In summary, the OECD WoE principles provide a practical tiered framework for testing and assessment of persistence. For example, multiple data for a given test method can be evaluated together. In cases where there is preexisting data, higher tier information (e.g., simulation data or calculations of Pov) should be more heavily weighted over lower tier information (e.g., QSAR or screening level tests) to characterize persistence. In silico tools such as QSARs and multimedia fate models, as well as read across, relevant physicochemical properties can be considered together using subject matter expertise (SME) and a WoE approach. When multiple values for the same metric are available for a given substance, the study designs and data interpretation should be examined closely to determine relevance of the findings as well as sources of variability so that the data can be appropriately considered a WoE evaluation. Types of information that would prove useful in assessing the overall reliability and quality of a study are included in Table 1.

Table 1. Persistence information for laboratory and field studies: Quality control and assessment measures Laboratory Fielda Experimental design External checks

Concentration

Replicates (appropriate)

Number of samples per interval relevance of environmental concentration

Appropriate positive (e.g., degradable substance) and negative controls (e.g., abiotic/sterile)

Redox status

Test substance characterization and purity Laboratory analysis of duplicate samples Robust analytical method Mode of substance application Sampling method and analysis method: Outlined and reproducible Analytics Internal checks (performed by the project field volunteers, staff, and laboratory)

Quality control and assessment measures

Calibration blank

Calibration standards, laboratory replicates

Detection/quantitation limit

Controls (e.g., shipping, extraction, etc.)

Source of biomass (soil, water, sediment sludge, etc.)

Description and characterization inoculum source or test medium (i.e., natural site described)

Biomass concentration and preparation

Viability verification throughout study

Table 2. Physicochemical properties and their impact on chemical fate Physicochemical property Impact on fate Molecular weight, water and octanol solubility, steric hindrance, ionizability, adsorptive properties, chemical form (e.g., liquid, solid, particle, etc.) Bioavailability Henry's constant in water, vapor pressure, Koa Air/water partitioning Ionizability, adsorptive properties, water solubility Soil or sediment/water partitioning Hydrolysable groups Potential modulation of degradation by environmental pH Photochemical reactivity Potential modulation of degradation by sunlight

Regardless, the concepts of gathering existing information, assessing reliability of information in a transparent and systematic manner, and interpreting results in a WoE framework are generally applicable to persistence assessments.