Intertek's Assurance in Action Podcast Network
Intertek's Assurance in Action Podcast Network
Cosmetic Industry: In Silico Methods
Today’s podcast is dedicated to In Silico methods including read-across and (Q)SAR approaches. The adoption of the European Cosmetic Regulation 1223/2009 has contributed to the development of alternative methods and tests to assess the hazards of new substances without animal testing. Join our French Intertek Assuris experts, Marie Duigou and Déborah Mitjans, as they talk about In Silico methods focusing on cosmetic ingredients, EU legislation & much more.
intertek.com
Follow us on- Intertek's Assurance In Action || Twitter || LinkedIn.
00:15 --> 01:09
Speaker 1
Hello everyone, welcome to the Assurance in Action podcast. Today’s podcast is dedicated to in silico methods including read-across and (Q)SAR approaches. The adoption of the European Cosmetic Regulation n° 1223/2009 has contributed to the development of alternative methods and tests to assess the hazards of new substances without animal testing. In silico models, in particular quantitative structure-activity relationship [or (Q)SAR] and read-across models, are a rapid and reliable alternative approach to in vivo and in vitro testing, and they play a key role in the development of toxicological profiles of new ingredients. My name is Marie and I am accompanied today by Déborah.
01:10 --> 01:32
Speaker 2
Hi everyone, and thanks Marie for the introduction. Marie and I are toxicologists within the Intertek Assuris team. As mentioned, the main topic of this podcast is in silico methods, and more precisely, we will focus on (Q)SAR and read-across approaches. First of all, Marie, can you tell us: what are in silico methods?
01:33 --> 02:08
Speaker 1
Sure. In silico toxicology includes a wide variety of computational tools, including databases, software, simulation tools, and modeling methods. In silico models and tools can predict the efficacy and toxicity of a substance, based on the principle that the characteristics of the substance’s molecular structure can be correlated with its physical and biological properties. The different in silico methods can be used in combination using a weight-of-evidence approach.
02:09 --> 02:17
Speaker 2
I see, thank you. This is a broad definition. What are the main methods used to characterize the hazards of a substance?
02:18 --> 03:48
Speaker 1
The three main methods are (Q)SAR, Expert Knowledge Models, and Read-Across.
Read-Across is a technique used to predict a specific endpoint of a substance by using experimental data from the same specific endpoint of one or several analog substances. Read-across studies are performed following the current guidelines set by the European Chemicals Agency and the Organization for Economic Co-operation and Development (the OECD) to meet the requirements for the Read-Across Assessment Framework.
(Q)SARs are mathematical models that can be used, in combination with an algorithm, to predict different endpoints based on the physical and structural characteristics of the molecules being tested (for example, molecular weight, number of rings, octanol-water partition coefficient, and etc.), also known as molecular descriptors.
Expert Knowledge Models use rules and data compiled by experts in models. The models allow us to group new substances by toxicity based on their known structures and the knowledge of the mode of action of certain chemical groups. Rules and decision trees can be established from these data to classify the new substance in an existing group, based on structural alerts associated with toxic activity.
03:49 --> 04:03
Speaker 2
I see. What’s interesting about in silico methods is that the tools concentrate so much data on the chemical substances that it is possible to obtain more or less precise information on any field. Do you agree with that Marie?
04:04 --> 04:59
Speaker 1
Yes, I totally agree. In practice, these methods can predict physicochemical properties, biological properties, and environmental fate. They can also predict toxicity endpoints such as skin sensitization, genotoxicity, and systemic toxicity. In silico models are in constant development, and improvements are regularly made to increase their reliability, thanks to the availability of large databases describing the properties and effects of chemicals, as well as powerful data-mining tools and various statistical algorithms. A number of in silico models and tools, such as the OECD Toolbox and VEGA, are currently available, and they cover a wide variety of chemical types and many key toxicological endpoints required for risk assessment.
05:00 --> 05:03
Speaker 2
So, what makes a model reliable?
05:04 --> 05:31
Speaker 1
To be reliable, a model must meet several criteria as defined by the OECD. The model must include, a defined endpoint; an unambiguous algorithm; a defined domain of applicability; appropriate measures of goodness-of-fit, robustness, and predictivity; and if possible a mechanistic interpretation
05:32 --> 05:38
Speaker 2
I see, but once we know that our model is reliable, how do we check the robustness of a prediction?
05:39 --> 06:02
Speaker 1
The OECD Toolbox and VEGA provide an applicability domain index to judge the accuracy of the prediction. In all cases, it is possible to strengthen a prediction by combining it with predictions from other models, such as from read-across, or from in vitro testing, in a weight-of-evidence approach.
06:03 --> 06:13
Speaker 2
So we can see that the robustness of a prediction depends a lot on the model, but is it possible to obtain a prediction on any substance?
06:14 --> 06:54
Speaker 1
In silico (Q)SAR tools work on the basis of the structure of the target substance. It is therefore not possible to use these methods to evaluate a substance whose composition is unknown or variable due to complex reactions or maybe biological materials. For example, ingredients of plant origin, such as botanical extracts, which are increasingly used in cosmetics, fall into this category. To evaluate their toxicity, it will be necessary to use in vitro methods or take into account their main phytochemical constituents one by one.
06:55 --> 07:11
Speaker 2
Thank you for your clarification. That is very helpful. We understand that many tools are put in place to validate these innovative methods, but what about the European regulations? At what level are these methods recognised by the agencies?
07:12 --> 08:32
Speaker 1
The use of in silico models is increasingly favoured in several regulation frameworks according to various fields of application. Read-across and in silico is encouraged by ECHA REACH Regulation to gain efficacy in the development and evaluation of new chemical substances, and these approaches are now widely proposed by registrants in a weight-of-evidence approach. According to the European Food Safety Authority guidance, a robust prediction using in silico methods with complete justification can be used to assess substances intended to be used in food contact materials. The Scientific Committee on Consumer Safety (or SCCS) indicates in their 11th revision of the Notes of Guidance for the testing of cosmetic ingredients and their safety evaluation, that (Q)SAR methods should be applied, when possible, to obtain a prediction of toxicity (like genotoxicity, skin sensitization, etc.) before considering any experimental test as part of the new methods "NAM" (New Approach Methodology) and "NGRA" (Next-Generation Risk Assessment).
08:33 --> 08:43
Speaker 2
If I understand correctly, in silico methods are thus used in many fields such as cosmetics, packaging, and also food contact materials.
08:44 --> 09:11
Speaker 1
Indeed, the fields of use for in silico methods are very broad. They range from defining the genotoxicity of a cosmetic substance to defining the metabolites of pharmaceutical substances and checking the mode of action of packaging impurities. Many solutions exist to help you to characterize the hazard of your substances. I invite you to contact us if you need some support.
09:12 --> 09:19
Speaker 2
Thank you, Marie, for this great conversation about in silico methods, and thanks to our listeners for joining us today.
09:19 --> 09:22
Speaker 1Thanks Déborah, and thanks for listening. Goodbye.