Difficult to Model Substances and Chemicals That Should Not be Profiled

There are a number of chemical classes that should not be run in the ECOSAR Class Program or types of compounds that are difficult to model. This is not an exhaustive list, but is intended to provide simple guidance to the user on limits of the tool.

Chemicals with Experimental Data: Chemicals with experimental data from a well conducted laboratory study should not be run through ECOSAR, unless it is for comparative purposes or a weight of evidence assessment. Experimental data should always be used in preference to estimations when identifying aquatic toxicity concerns.

Inorganic Chemicals: The computerized estimation methods currently programmed into the ECOSAR Class Program were designed and developed for organic chemicals. Inorganic chemicals will not provide reliable results and should not be profiled. It includes neutral species such as titanium dioxide (TiO₂) and inorganic salts, such as sodium chloride (NaCl) or potassium permanganate (KMnO₄). This class of chemicals also includes organo-metallic chemicals (chemicals that contain carbon bonded to a metal species). There is a current lack of good data and knowledge rules regarding toxicity pathways for these types of compounds, and the traditional estimation techniques used for organics have in some cases been found to be inappropriate for metal containing molecules. It is also difficult to distinguish oxidation states using the molecular descriptors employed for organics, which may play an important role in estimating toxicity for metals.

If a user enters a structure into ECOSAR which contains a metal atom for which ECOSAR is inappropriate, the model will provide a warning to the user.

Hydrolytically Unstable or Highly Reactive Chemicals: Data or information on a chemical’s degradation rate can determine if other chemical degradates should be considered during the course of an ecotoxicity assessment. Guidance provided within the context of the U.S. EPA New Chemicals Program states that if a chemicals half-life in water is less than one hour, it is predominantly the degradation products that enter the aquatic environment and the focus of the ecotoxicity assessment should be on the degradates, not the parent. If the half-life is greater than one hour, but less than 14 days, then both the intact parent chemical and its degradates should be reviewed. If the half-life is greater than 14 days, then only the parent chemical is generally assessed. This 14 day criterion is based on the average residence time for a chemical to move through a waste water treatment facility before being released to surface water where it may be toxic to the aquatic life. The form a chemical takes after this 14-day window is important to consider when determining what the aquatic system will be exposed to (parent or degradate). If a user suspects a query chemical may be highly reactive, the user may choose to run the ECOSAR model on the degradation products instead of the parent product. For example, peroxides would be expected to degrade to a pair of corresponding alcohols (or only a single alcohol for asymmetrical peroxide). These alcohols could then be profiled to determine their potential toxicity.

Complex Organic Salts: The properties and toxicity of a small number of organic salts are well documented in the environmental literature such as Sodium (Na) salts, Potassium (K) salts, Ammonium (NH4+) salts, and Lithium (Li) salts. Simple salts can be entered into the system and modeled by neutralizing the compounds (remove the salt, replace with a hydrogen atom) as the salts are typically treated as spectator ions. The user must of course consider compounds and dissociation at environmental conditions. Compounds with more complex organic cation and anion are difficult to model because the organic salt may in fact play a larger role depending on dissociation products. In this case you must make a more careful inspection of anion, cation, and dissociation. Some users may choose to run as single compound (neutralized with both cation and anion attached), or run as separate individual compounds (dissociated with no charge). There is no specified procedure in these cases; it is done on a case-by-case basis.

High Molecular Weight Compounds (MW >1000): Polymers and chemicals with a molecular weight greater than 1,000 should not be profiled using this tool, as the ECOSAR equations were not developed using these types of compounds. However, many polymers may be made up of dimers, trimers, and oligomers that have a molecular weight of less than 1,000. These smaller molecules often contain the same components of the larger polymers, and, therefore, could be run through the ECOSAR model when performing an aquatic toxicity assessment. For chemicals with MW > 1000 and various types of polymers, please see the Interpretive Assistance Document for Polymers on the U.S. EPA’s Sustainable Futures website: http://www.epa.gov/oppt/sf/meetings/train.htm#materials

Isomers: Three dimensional molecular properties or molecular conformation can be important as it relates to absorption, binding, and resulting toxicity potential of a chemical. Some QSAR models are unable to account for these three dimensional characteristics that in some cases can be important considerations since they can influence toxicokenetic (PBPK) processes. Often QSAR models do not distinguish between steroisomers, optical isomers, tautomers, or specific conformations because they are built using simple one or two-dimensional descriptors only, as is the case with the ECOSAR model. More advanced QSAR techniques and experts systems may be able to distinguish isomers either through advanced calculations or overlaid expert rules, but the user often sacrifices speed when using these tools.

Mixtures: ECOSAR does not have the ability to simultaneously assess multiple chemicals and all potential synergistic effects of that mixture. The model uses more simplistic structure-activity relationships based on a single, discrete chemical structure as its input. There are many CAS numbers that actually represent isomer mixtures, polymers (containing residual monomers), and unknown or variable composition substances. In general, mixtures cannot be run through QSAR models. If the chemical to be profiled is a mixture of discrete organic substances, then each substance can be run through the model separately and the result can be compared and contrasted. If there is one component of a mixture that predominates, then it may be used to represent the entire mixture (i.e., a representative structure can be entered).

Chemicals with Unknown or Variable Composition: If the material has a variable composition, (such as oligomers, natural fats, or a product mixture that changes composition depending on the reaction conditions), then the results provided by models where a representative structure was used may not accurately reflect the true nature of the commercial product. Because of the complexity of structural identification, the assessor may need to complete a review of the CAS, name, reaction mechanism, and representative structure(s) to determine appropriateness of the structural input used to generate the model estimates. If this procedure is performed, the results should be interpreted with caution, as other components of the product may possess significantly different toxicities