Mapping Chemical Respiratory Sensitization: How Useful Are Our Current Computational Tools?

Abstract

Chemical respiratory sensitization is an immunological process that manifests clinically mostly as occupational asthma and is responsible for 1 in 6 cases of adult asthma, although this may be an underestimate of the prevalence, as it is under-diagnosed. Occupational asthma results in unemployment for roughly one-third of those affected due to severe health issues. Despite its high prevalence, chemical respiratory sensitization is difficult to predict, as there are currently no validated models and the mechanisms are not entirely understood, creating a significant challenge for regulatory bodies and industry alike. The Adverse Outcome Pathway (AOP) for respiratory sensitization is currently incomplete. However, some key events have been identified, and there is overlap with the comparatively well-characterized AOP for dermal sensitization. Because of this, and the fact that dermal sensitization is often assessed by in vivo, in chemico, or in silico methods, regulatory bodies are defaulting to the dermal sensitization status of chemicals as a proxy for respiratory sensitization status when evaluating chemical safety. We identified a data set of known human respiratory sensitizers, which we used to investigate the accuracy of a structural alert model, Toxtree, designed for skin sensitization and the Centre for Occupational and Environmental Health (COEH)’s model, a model developed specifically for occupational asthma. While both models had a reasonable level of accuracy, the COEH model achieved the highest balanced accuracy at 76%; when the models agreed, the overall accuracy was 87%. There were important differences between the models: Toxtree had superior performance for some structural alerts and some categories of well-characterized skin sensitizers, while the COEH model had high accuracy in identifying sensitizers that lacked identified skin sensitization reactivity domains. Overall, both models achieved respectable accuracy. However, neither model addresses potency, which, along with data quality, remains a hurdle, and the field must prioritize these issues to move forward.

Computer-Aided Discovery and Redesign for Respiratory Sensitization: A Tiered Mechanistic Model to Deliver Robust Performance Across a Diverse Chemical Space

“…Thus, it is reasonable to propose that for hexachlorophene, the observational nature of the experimental data is creating a level of uncertainty that cannot be resolved with computational modeling. 5 Similarly, 2-methyl-3,5-dinitrobenzamide is difficult to explain as a direct respiratory sensitizer, when picric acid and trinitrotoluene (Figure 7b, false positives) are presumably nonsensitizing. The compound could be a photoallergen, forming a nitro radical anion on photolysis that is coplanar with the aromatic ring; the radical anion radical can then be metabolized into a reactive intermediate.…”

“…In surveying the landscape of existing in vivo, in vitro, and in silico approaches, no single model convincingly demonstrates the ability to predict respiratory sensitization with a high degree of confidence, leading to proposals of (difficult-tovalidate) consensus-type schemes as workable solutions. 4,5 Owing to its premier status in evaluating dermal sensitization in quantitative risk assessment, regulatory bodies often default to the murine local lymph node assay (LLNA) as a useful surrogate for a respiratory-sensitization model. This strategy is based on a notion that a chemical with a negative response in the LLNA usually lacks the ability to induce both skin and respiratory sensitization.…”

“…Data issues aside, the common drawback of existing in silico models for respiratory sensitization is their (over)­reliance on alerts and structural similarities rather than chemical interactions in KEs. ,, While mechanistically derived alerts can provide reasonable predictivity within existing knowledge, as shown repeatedly for the dermal sensitization end point, these approaches often fail when venturing into a new chemical space. , This is because high predictivity demands high specificity of the alert, which increases false-negative rates for new chemistries, and because the chemical environmental surrounding an alert can both impede and promote its reactivity, as notable for heavily functionalized pharmaceuticals, for example. One strategy for overcoming this challenge is to use structural alerts as a “permissive mechanistic screen”, which can be supplemented by an expert review of new chemistries, and to rely on more rigorous modeling of molecular initiating events (MIEs), such as covalent protein haptenation, to determine sensitization potential or potency .…”

“…In surveying the landscape of existing in vivo, in vitro , and in silico approaches, no single model convincingly demonstrates the ability to predict respiratory sensitization with a high degree of confidence, leading to proposals of (difficult-to-validate) consensus-type schemes as workable solutions. , Owing to its premier status in evaluating dermal sensitization in quantitative risk assessment, regulatory bodies often default to the murine local lymph node assay (LLNA) as a useful surrogate for a respiratory-sensitization model. This strategy is based on a notion that a chemical with a negative response in the LLNA usually lacks the ability to induce both skin and respiratory sensitization. − However, there are exceptions, in part because this assumption ignores the effect of absorption kinetics on the toxic response, where crossing the lung mucosa is driven by different physicochemical properties than permeation through the stratum corneum of skin (though some chemicals can also sensitize the respiratory tract via the dermal route) .…”

Introduction to Special Issue: Computational Toxicology

“…Both of these groups provide these resources in web-based open access formats with user-friendly interfaces. Golden et al discuss a related end point, respiratory sensitization, where chemicals trigger an immune response via inhalation exposure routes, leading to occupational asthma. While models for skin sensitization can adequately predict some respiratory sensitizers, there is no standardized testing strategy or reliable reference data to train in silico models, and work to predict this end point would benefit from further curation efforts and AOP development.…”