Hello to everyone joining us and welcome to SenzaGen's webinar. My name is Tina Lawesson. I'm handling marketing and communications at SenzaGen, and I will be your host today. Today's talk is a recap on SenzaGen's scientific session at the SOT meeting in San Diego last month. This is for you who missed it and for you who wants to listen to it again. We will start with a 50-minute recorded presentation, followed by a live Q&A session. Regarding the Q&A, you can submit your questions at any time during the presentations. You will just have to use the GoToWebinar panel and write your question in the questions chat box. The chat box is located on the right side of your screen. You should be able to find it there. Thank you for joining us today. We are now ready to start the recorded presentation.
Here we go.
Good afternoon, everyone, and welcome to today's webinar. My name is Andy Forreryd, Scientific Liaison Manager, SenzaGen, and I will take you through today's session, which is a recap from our industry hosted session at the recent SOT meeting. During the webinar, we will also have the opportunity to listen to presentations from two invited guest speakers, Dr. Christopher Choi from Takasago, and Dr. Karla Lienau from Sonova. We will together present a selection of case studies and user cases where the GARD technology platform have been used in order to fill some of the remaining data gaps associated with in vitro skin sensitization testing, with a specific focus on applicability domain and quantitative potency assessment. This will be a pre-recorded webinar ending with a live Q&A session, where we will be available to answer your questions regarding the GARD technology platform and the different applications.
Thank you very much for joining today's session. I hope you enjoy the presentation, and we are looking forward to a really interesting Q&A session. Starting with a brief look at today's agenda, I would like to start this presentation with an introduction to the field of skin sensitization testing and to the GARD technology platform. Then present novel case studies on the testing of metals, agrochemical formulations, and fragrance materials generated in collaboration with industrial collaborators. Our first guest speakers of today's session is Dr. Christopher Choi, Vice President of Corporate Safety and Regulatory Affairs at Takasago. Dr. Choi will present a use case on the use of the GARD assay to assess the skin sensitizing potency in Takasago's fragrance ingredient safety program. Our second guest speakers of today's session is Dr. Karla Lienau, Biological Safety Specialist at Sonova.
Dr. Karla Lienau will present a user case on how the GARD assay have been used for quantitative potency assessment of skin sensitizers during the development of novel materials intended for use within a medical device. This presentation will then end with a Q&A session, and we are, of course, looking forward to really interesting discussion. Starting with a very brief overview of the mechanism of skin sensitization, and I of course, appreciate that the majority of you are already well familiar with this process. The development of skin sensitization is dependent on the activation of an adaptive immune response. The first step in this process is essentially a chemical that binds to the endogenous proteins in the skin in order to generate a hapten-protein complex.
This complex is then recognized by dendritic cells, which in response to different type of inflammatory mediators, become activated and migrates to the draining lymph node, where they activate allergen specific T cells and give rise to an immunological memory. Each time that you're then exposed to the same chemical, you will have a rapid T cell proliferation, which will eventually give rise to an inflammatory response in the skin, which is characterized as allergic contact dermatitis. The mechanistic information from the previous slide are often summarized in what is known as the adverse outcome pathway for skin sensitization, which highlights the four most important key events that are required for the development of skin sensitization. The development of different in vitro test methods for the prediction of skin sensitization has leveraged on this mechanistic information.
Today, there is a total of eight in vitro methods that have been formally validated and incorporated into the globally accepted test guidelines by the OECD. These methods all target individual key events in the adverse outcome pathway for skin sensitization. This presentation will focus exclusively on the genomic allergen rapid detection assay, for which there is now also a draft OECD test guideline available. I think it deserves to be mentioned that in a regulatory context, none of these OECD in vitro test guidelines are currently considered a sufficient standalone replacement for the animal data. Instead, these assays should be used together, either within a weight of evidence approach or within a defined approach.
With the recent release of the OECD Test Guideline 497, there are today three such defined approaches that have reached regulatory acceptance at OECD level. These are the two out of three method for binary hazard classifications. There is two different versions of the ITS for potency assessment, according to the subcategories defined by Globally Harmonized System. There are currently no defined approach that are accepted for quantitative assessment of relative skin sensitizing potency for use within a risk assessment procedure. The current OECD validated assays, and of course, also the defined approaches are, of course, really useful in many situations. I think it's also important to acknowledge the fact that they are associated with certain limitations.
The aim of this slide is to highlight some of the challenges and data gaps that remain to be addressed by novel in vitro assays, such as the GARD technology platform. In terms of applicability domain, the OECD assays have been validated on a rather narrow subset of the chemical space, or there may also be a lack of data to support the inclusion of certain type of chemicals into the applicability domain of the individual assay. In case study one and two, I would present data to support the inclusion of metals and also agrochemical formulations into the applicability domain of the GARD assay.
The next gap I would like to highlight in this presentation is related to the potential of these assays to provide a quantitative assessment of relative sensitizing potency on a continuous scale that can be used for a risk assessment procedure. The potential to use the GARD assay for this purpose will be presented by our invited guest speakers in today's presentation, but they will also be covered in case study number three in my presentation. The final gap I would like to highlight today in the presentation is related to the potential to use this in vitro assays also for biocompatibility testing of medical devices. For the endpoint of skin sensitization, there is currently no validated in vitro assay available for this purpose.
I will briefly cover the GARDskin medical device application in my presentation, and we will also later hear from Dr. Lienau on how this method has been used in practice by the industry. With that, I would now like to provide you also with an introduction to the GARD technology platform and describe some of the unique technological feature of this assay, and also try to demonstrate how this assay brings novel elements to the field of regulatory toxicology by using a combination of high content gene expression data and machine learning. The GARD technology platform currently offers a selection of different in vitro assays for the endpoint of skin sensitization that can be selected either dependent on the amount of potency information that is required or based on the properties of the test materials.
The available assays include the GARDskin assay for hazard assessment of skin sensitizers, the GARD potency assay for potency sub-categorization as weak and strong sensitizers, according to the CLP regulation, or the GARDskin dose response assay, which is a modification of the standard GARDskin protocol in order to enable for more quantitative assessment of relative skin sensitizing potency on a continuous scale for use within a risk assessment procedure. Finally, there is also the GARDskin medical device assay that can be used for testing of solid materials and medical devices, in compliance to the ISO standard for biocompatibility testing of medical devices. This slide provides an overview of the general principles of the GARD technology. GARD is based on a dendritic cell line, therefore targets Key Event 3 in the adverse outcome pathway for skin sensitization.
Similar to many other toxicological assays, the GARD assay also monitors how these cells react following the exposure to a test chemical and then utilize this information in order to classify the unknown test substance as either a skin sensitizer or a non-sensitizer. The unique feature of the GARD assay is that this cellular response is not only monitored using one or two biomarkers, but instead a gene expression across a larger set of genes are being monitored in order to arrive at a more holistic result of how these cells react to the external stimuli. In the GARDskin assay, for example, a total of 200 genes are being monitored and used for the final classification.
In order to get a better understanding of the biological relevance of the 200 genes in the GARDskin prediction signature that are being used for the final prediction and how this actually relates to the biology behind skin sensitization, I have here mapped the genes and their associated toxicity pathways to the adverse outcome pathway of skin sensitization. Here we can see that the genes monitor cellular events that are associated with Key Event 3, dendritic cell activation, such as, for example, the costimulatory molecule CD86, and also different type of Toll-like receptors. We can also see that genes are being monitored that are classically associated with Key Event number 2, such as the protection against the oxidative stress and the generation of a proinflammatory response.
Overall, the gene expression signature monitors a large set of genes that are involved in immunologically associated pathways that are relevant to the sensitization process. The GARD workflow can essentially be summarized in six steps, and stimulations are performed at one single concentration, which is mainly dependent on the solubility and the cytotoxicity properties of the test materials. Once this concentration has been identified, stimulations are performed again, and cells are incubated with the chemical for 24 hours. After that, you isolate all the RNA materials, and you measure the gene expression of the 200 genes in the predictive biomarker signature. In the final two steps, a machine learning algorithm is used in order to compare the gene expression profile induced by the unknown test item to the gene expression profile induced by sensitizers and non-sensitizers in our training data.
The interpretation of the results are very simple. The classification threshold that are being used is zero and a positive decision value from the algorithm is classified the unknown test item as a skin sensitizer and a negative value classified as a non-sensitizer. Having now introduced you to the main concept behind the GARD technology platform, I also want to provide you with a short introduction to the GARD potency assay and describe how this assay can be used in order to further sub-categorize identified skin sensitizers based on the two potency categories defined in the GHS CLP system as strong and weak sensitizers.
The GARD potency assay is based on exactly the same protocol as the previously induced GARDskin assay, but it monitors a complementary set of 51 genes that can be used in order to further sub-categorize identified skin sensitizers as either strong or weak sensitizers. The intended use of the GARD potency signature is within a defined approach where the skin sensitizers are first identified in the GARDskin assay and then further sub-categorized as either strong or weak sensitizers using the GARD potency assay. The performance of the GARDskin and the potency assay have been validated in a blinded ring trial, including three participating laboratories, that was performed in full compliance to the OECD documents. This slide shows the data from the GARDskin and the GARD potency assay, with the data from the GARDskin assay in the table here also, below.
As you can see from this figure, the assays demonstrated a high predicted performance, and the results were also highly reproducible between the participating laboratories. The EURL ECVAM Scientific Advisory Committee, the ESAC, has also conducted an independent review of the scientific validity of the GARD assays. They concluded that some additional data is required for the GARD potency assay, mainly to demonstrate on the reproducibility, but that the GARDskin assay is now ready to progress to further consideration by the OECD for test guideline development. There is now also a draft OECD guideline available for this assay that are up for approval at OECD level.
This ESAC statement on the GARD assay actually constitute a landmark opinion, since it's the first time a machine learning algorithm has been independently reviewed for the use within the field of regulatory toxicology, and thereby setting the precedent for how to evaluate this type of assays in the future. Having now introduced both the GARDskin and the GARD potency assay, as well as the main technological features of the platform, I would now like to move on to introduce also the GARDskin dose-response assay, which can be used in order to derive a more, quantitative assessment of relative skin sensitizing potency on a continuous scale for use within a risk assessment procedure. The GARDskin dose-response assay is based on the validated protocols of the GARDskin assay.
Instead of performing testing only at one single fixed concentration, testing in the GARDskin Dose-Response assay is performed within a titrated range of different concentration. The standard GARDskin protocols are used in order to generate classifications or decision values for each of the tested concentration, and the resulting data is then used in order to generate a dose response curve by plotting the output from the machine learning algorithm versus the input concentration. You can see as an example on this in the figure here to the right. The endpoint or the output in this model is not only a binary hazard classification, but here we are aiming to identify the lowest concentration required to exceed the binary classification threshold of zero and thereby induce a positive classification in the GARDskin assay. We refer to this concentration at the CDV0 value.
For those of you that are well familiar with the local lymph node assay, you can see that this is a complete analog to how the local lymph node assay actually works. This is some data that was also recently published in Nature Scientific Reports. These figures are from an initial proof of concept study, including a total of 29 chemicals which were published in the Scientific Reports publication last year, aiming to investigate the correlation between the experimentally derived cDV0 values from the GARDskin Dose-Response assay and the continuous EC3 values from the local lymph node assay in the figure to the left and the human NOEL values in the figure to the right.
As you can see in the figures, these experimentally derived cDV0 values from the GARDskin Dose-Response assay were significantly and strongly correlated both to LLNA EC3 values and the human NOEL values. Based on the observed correlation between the cDV0 values and potency, we have also proposed a protocol for potency prediction of unknown test substances based on linear regression models using the experimentally derived cDV0 concentrations as points of departure. I tried to illustrate this procedure here using the compound resorcinol, but it would of course work equally well for unknown test materials.
In the first step, you would basically perform the GARDskin Dose-Response protocol in order to generate a dose-response curve and identify the cDV0 concentration. In the second step, which is here illustrated in the figure to the right, the derived cDV0 value for the compound is then used as input into the linear regression model, which is used in order to predict a corresponding LLNA EC3 value or a human NOEL value. We have of course evaluated the precision of this approach using a variety of different external and blinded test data sets, and some of this data will also be presented in the case studies later on in this presentation.
Also finally, before I move into the different case studies, I would also like to introduce the GARDskin Medical Device application, which enable for skin sensitization testing of solid materials and medical devices in compliance to the available ISO standard. The GARDskin Medical Device assay is based on the validated protocols of GARDskin, but the protocol has been modified to include a sample extraction procedure using both a polar and a non-polar vehicle in compliance to the ISO standard. In the figure here, you can see the results from an initial proof of concept study that were designed in order to evaluate the functionality of this protocol for testing of medical devices. In this project, we evaluated a set of polymeric materials that were spiked with different sensitizers, and in this case, used as proxies for medical devices.
The extractions here were performed in both a polar and a non-polar solvents, which were either an olive oil or a sesame oil, and we of course classified the leachables in all of the solvents using the conventional GARDskin assay. As you can see in the figure, the spiked materials were correctly classified as skin sensitizers in all the different vehicles, while still avoiding false positive classifications from the medical-grade control materials. This study demonstrated the compatibility of the GARD assay, both to polar and non-polar extraction vehicles, and also demonstrated that the assay was sensitive enough to detect the presence of sensitizers in the extracts. That essentially concludes the first phase of this presentation, the introduction to the different GARD assays.
In the next part of this presentation, I would like to highlight a few case studies to demonstrate how these assays can also be used in practice. This data have been generated in collaboration with different, industrial collaborators, and the first case is a collaboration between SenzaGen and Johnson Matthey to evaluate the applicability domain of GARDskin for assessment of metals. The second case study is a collaboration between SenzaGen and Corteva Agriscience to evaluate the applicability domain of GARDskin and GARDpotency for assessment of agrochemical formulations. The final case study is a collaboration between SenzaGen and International Flavors & Fragrances to evaluate the applicability of the GARDskin Dose-Response assay for quantitative assessment of skin sensitizing potency of fragrance materials.
The aim of the first case study was to generate experimental data to support the inclusion of metals into the applicability domain of the GARDskin assay. Metals are generally considered as either outside the applicability domain of the current NAM-based approaches, or there is simply not enough data to demonstrate the applicability. In this specific case study, a total of 13 metals were provided by the sponsor, and testing were performed according to the draft OECD test guideline and the standard protocols, with only a minor adjustment regarding the selection of solvent in order to increase the solubility and to ensure the test item stability. For example, we did not use DMSO in this study because it's known to interact with the platinum and then potentially modify also the toxicity profile of the test material.
The result from the testing are presented here in the figure, where you have the expected non-sensitizers to the left in the figure and the sensitizers to the right based on available reference data. On the y-axis, you then have the GARDskin classifications. Just to remind you again, a positive value is classified as a skin sensitizer and negative value is classified as a non-sensitizer. Overall, a high predicted performance was reported for this testing with only one false positive classification. In conclusion, the performance values that were reported in this study for metals were very similar to previously reported numbers for other low molecular weight organic molecules, and subsequently provide support for the inclusion of metals into the applicability domain of the GARD assay.
The second case study was a collaboration between SenzaGen and Corteva Agriscience to evaluate the applicability of the GARDskin and the GARDpotency assay for assessment of agrochemical formulations. There is today very little data available in the scientific literature regarding the performance of the OECD-validated assays for assessment of multi-constituents and formulations. Most of these assays also been validated using neat materials or single ingredients. In this specific case, a total of 20 liquid-based agrochemical formulations comprising a mixture of water and solvent-based formulations were provided by Corteva. Testing were performed in a blinded fashion according to the standard protocols. Since we did not have any molecular weight for the formulations, the concentrations were established based on an approximated molecular weight of 400 g per mole, which also correspond to the molecular weight of the active ingredients.
Testing were then initially performed in the GARDskin assay, and for formulations that were classified as skin sensitizers, they were also further evaluated in the GARDpotency assay in order to derive a CLP potency classification. This slide summarizes the results from the testing. A total of 15 formulations of the 20 were correctly predicted, with one false negative and four false positive classifications. In terms of GARD potencies, six out of the eight category 1B were correctly classified, with one over-prediction and one under-prediction. By looking a bit more in detail at the LLNA reference data for the misclassifications in GARD, which I have here highlighted in yellow, we can also see that the misclassification were generally associated with borderline in vivo results.
Either having a stimulation index very close to the classification threshold of three, or having a really high EC3 value, which may also help to explain the lack of the concordance for these specific substances. In conclusion, the GARDskin and the GARD potency assay showed very promising concordance to the available reference data for assessment also of these type of complex phytochemical formulations, and the misclassification were generally attributable to borderline in vivo results. Here it's important to bear in mind that the local lymph node assay may not always be representative of the situation in humans. The final case study was a collaboration between SenzaGen and IFF to evaluate the GARDskin Dose-Response assay for quantitative assessment of skin sensitizing potency of fragrance materials.
This type of quantitative prediction on a continuous scale is required in order to define a point of departure that can be used for risk assessments. In this specific study, a total of 12 fragrance materials were kindly provided by IFF, and these also included slightly more complex samples, such as a UVCB and a multi-constituent. In this project, testing were performed in a completely blinded manner according to the standard protocols. We used derived cDV0 values in order to predict a corresponding Local Lymph Node Assay EC3 value and a human NOEL value. In the figure set to the right, you can see two representative dose-response curves from the study. On top, you can see the dose-response curve for one of the sensitizers in the project.
You can see here also that the shaded area represents the confidence interval in the predictions. In the figure, at the bottom, you can see that there is basically no response at all, and this corresponds very well to the substance also being a non-sensitizer. You can see all of the classifications were below the classification threshold required for classifying these substances as skin sensitizer. This slide summarizes the results from the binary hazard classifications, as well as the CLP potency predictions based on the predicted human NOEL values from the GARDskin Dose-Response model. Two materials were removed from the calculations since the reference data were non-conclusive from these materials. For the hazard classification, nine out of the 10 materials were correctly predicted, with one false negative classification.
For the potency prediction, we used a human NOEL value of 500 mcg/cm² as a cutoff for assignment of the CLP categories. In the figure here to the right, you can basically see the predicted human NOEL values from the GARD assay normalized to the classification threshold. A value less than one indicate Category 1A prediction and a value above one indicate a Category 1B classification. Based on the results from the potency assignments, all of the Category 1A were correctly predicted, and four out of the six Category 1B were correctly predicted, with one under-prediction and one over-prediction. The next step, and also potentially the most important step, was to evaluate how the predictions, potency predictions from the GARDskin Dose-Response assay correlated to the available potency references on a continuous scale.
As you can see in the figure to the left, the predicted human NOEL values from the GARDskin Dose-Response assay correlated extremely well with the human NOEL reference data, with only one material being predicted at slightly more potent sensitizer in the GARD assay than the available human data suggest. In the figure to the right, you can see that also the predicted Local Lymph Node Assay EC3 values from the GARDskin Dose-Response assay correlated quite well with the reference values, except for three materials, which were predicted as slightly more potent sensitizers in the GARDskin Dose-Response assay than the reference data suggested. Finally, also to conclude and summarize some of the results from this study, starting with the discrete classifications, the GARDskin Dose-Response assay generated accurate predictions for both hazard and CLP potency classifications.
If we look at a slightly more granular scale and evaluate also the continuous prediction, we can see that the GARDskin Dose-Response assay predicted the human NOEL values with a really high precision. We can also see that the GARDskin Dose-Response predicted the Local Lymph Node Assay EC3 values with high precision for most of the materials, with the deviating value mainly attributable to differences between the Local Lymph Node Assay values and the human reference data. Overall, based on the data generated in this study, but also based on in-house generated data, the GARDskin Dose-Response assay appears to be functional across a variety of different materials, including also multi-constituents, UVCBs and quite hydrophobic materials.
That concludes my part of the presentation. Without any further ado, I would like to introduce the first guest speakers of today's webinar, which is Dr. Christopher Choi, Vice President for Corporate Safety and Regulatory Affairs at Takasago. Dr. Choi is a board-certified toxicologist, having worked in a variety of different industries, mainly involved in toxicological testing and risk assessment to ensure human health and environmental safety. In his current position, Dr. Choi oversees the safety program for novel fragrance ingredients and have been involved in various aspects of adapting more non-animal-based testing methods into the safety program at Takasago. Christopher will present the use of the GARD assays to assess the skin sensitizing potency in Takasago's fragrance ingredient safety program.
Thank you for the introduction, Andy. My name is Chris Choi, and I'm the Vice President of Corporate Safety and Regulatory Affairs for Takasago International Corporation. Today, I'll be going over the use of the GARD assay to assess skin sensitization potency in Takasago's fragrance ingredient safety program. I've structured my presentation to first go over the previous safety testing paradigms and the challenges that we face with in vitro alternative methods to assess for skin sensitization. Then I'll go into alternative approaches incorporating the GARD potency assay into our safety testing program. Lastly, I will go over two test cases where we've taken two novel fragrance ingredients and evaluated through the alternative approaches. In looking at the standard testing paradigm based on current regulatory requirements, the novel fragrance ingredients, skin sensitization endpoint is required for both the European ECHA chemical legislation as well as the U.S. EPA TSCA registration.
Both regulatory requirements are aimed to ensure safety in both occupational and consumer product exposures. This endpoint can be satisfied with either an in vivo assay, which includes the Local Lymph Node Assay or the guinea pig maximization test. It can be accomplished with a battery of in vitro assays, the two out of three method. Now, the drawbacks with in vivo assay is that animal testing is required. The drawback with in vitro assay is that there's a lack of potency information with in vitro assays when positive or equivocal findings are observed. In addition, it also tends to be a bit more expensive. We've incorporated the GARD potency assay into our alternative approaches. The GARDskin Dose-Response assay utilizes the SenzaGen cell lines to generate a dose-response curve.
The Dose-Response curves are based on the lowest exposure concentration to test article required to elicit a positive classification which is identified as a cDV0 value. There's a significant correlation between the GARDskin Dose-Response cDV0 value and the Local Lymph Node EC3 as well as the human NESIL values. We would also incorporate the DPRA assay as well as the KeratinoSens assay, if required, to evaluate the two out of three approach. In addition, we would also look at the weight of evidence approach by looking at the skin sensitization data of similar chemicals within the chemical space, as well as the components of the chemicals. Then we could follow up with a confirmatory HRIPT assay to confirm that there is no skin sensitization concerns. We've identified two novel fragrance ingredients, for further development and safety testing under this new testing paradigm.
For fragrance ingredient one, the molecular weight is roughly around 160 Dalton as they expect it to penetrate the skin. It also has a furan ring. It's classified as a Cramer class III. In looking at the in silico assays, it was considered a non-sensitizer with the Derek Nexus program. There was no skin sensitization domains identified with Toxtree. Using the OECD Toolbox, no skin sensitization properties were identified. For fragrance ingredient number two, the molecular weight is roughly about 170 Daltons. It's also expected to penetrate skin. It has an oxime structure. It has a Cramer class classification three. It was found to have equivocal concern for skin sensitization with the alkoxy part with the Derek program.
No skin sensitization domains were identified with the Toxtree, and then with the OECD Toolbox also, no skin sensitization properties were identified. When we took the first fragrance ingredient through the testing paradigm, the following GARDskin Dose-Response results were obtained. The maximum concentration tested was identified at 100 micromolars, which resulted in a cell viability of greater than 95%. It was classified as a sensitizer based on the results of the study, and a cDV0 value of 18.4 micromolar was obtained. Now, this corresponded with an estimated LLNA EC3 value of approximately 1.93%, an estimated human NESIL value of 659 mcg/ cm².
When we took this fragrance ingredient through the DPRA assay, it was found that it was not categorized as subcategory A, and the peptide depletion at maximum was about 20.4% observed at 1,440 minutes. As the two assays did not correspond to a true positive, we're taking this ingredient through a KeratinoSens assay to determine whether there is a true sensitization potential using the three out of three method. For fragrance ingredient number two, the maximum concentration tested was identified at 500 micromolar, which resulted in a cell viability of greater than 91%. It was classified as a sensitizer based on the results of the study, and the cDV0 value of 296 micromolar was obtained.
This corresponded to an estimated LLNA EC3 value of 27.8%, an estimated human NESIL value of 16,600 mcg/cm² . This was also taken through the KeratinoSens assay, and the results indicated that it was not categorized as subcategory 1A, and the peptide depletion was observed at 30 minutes to be approximately 1%. Likewise, with the previous ingredient, a KeratinoSens assay is in preparation to determine whether it has any skin sensitization potential using the two out of three method. In conclusion, the results from both compounds would suggest that there may be a potential for skin sensitization. At least one out of the two assay would demonstrate a clear positive, and a third test will be required to make a final decision if we're taking the two out of three approach.
Now, if the third test comes back negative, we can either take the non-reactive DST approach to determine the maximum use concentration for these fragrance ingredients in a final consumer product. We can conduct a confirmatory HRIPT test, of course, the concentration to be determined in the future, to determine a safe maximum use concentration that we can utilize for the fragrance ingredient. Now, if the third test comes back positive for fragrance ingredient number one, we've looked at similar compounds with furan ring, and there's HRIPT data that would suggest that at 1.25%-2.5%, which corresponds to roughly NESIL values of 860-1700 mcg/cm² , that it would not result in any concern for skin sensitization.
We will potentially conduct a confirmatory HRIPT testing at 600 mcg/cm² , which is very close to the data that was observed from the GARDskin Dose-Response assay. For fragrance ingredient number two, similar compounds with oxime structures were demonstrated to have HRIPT data with testing between 20%-100% concentration. This corresponds to roughly NESIL values of 13,000 mcg/cm²-69,000 mcg/cm² without any evidence of skin sensitization. We will conduct a confirmatory HRIPT testing at about 16,000 mcg/cm² , which also corresponds very closely with the GARDskin Dose-Response assay. With that, I'd like to finish my talk and provide a big thanks to the Takasago team, Mr. Satoshi Sasaki, Reshma Ramdhany , as well as the SenzaGen team, for the support they provided for these two materials. Thank you.
Thank you very much, Christopher, for that really nice presentation, and thank you for introducing us to the safety testing program at Takasago. I would now like to introduce our next guest speaker of today's session, Dr. Karla Lienau, Biological Safety Specialist and Research Engineer at Sonova. Dr. Lienau has a background of chemistry, and she works as a Research Engineer to develop and evaluate new materials, coatings, and technologies with a specific focus on custom hearing aids. She also conducts biological evaluations according to the ISO standard for evaluation of medical devices of class II A. The title of Karla's talk is In Vitro Methods for Quantitative Potency Assessment of Skin Sensitizers during Development of Novel Materials for Intended Use in Medical Devices.
Thank you, Andy, for the introduction. As Andy mentioned, I will talk about the application of the dose-response assay as an in vitro method for quantitative potency assessment of skin sensitizers during the development of novel materials for the intended use in medical devices. First, I want to introduce you to Sonova. Who is Sonova? What are our products? I will talk about custom product materials for which we did this dose-response assay. In the end, I will talk about GARDskin Medical Device use cases. At Sonova, we envision a world where everyone enjoys the delight of hearing and therefore lives life without limitations. Sonova grew from a small backyard workshop in Switzerland to a global company. To give you an idea of our size, I wanted to show you some key numbers.
Last year, business year 2021, we had sales of CHF 2.6 billion, over 14,000 employees, and a net profit of CHF 585 million. Operates through its core businesses, hearing instruments, cochlear implants, and audiological care with the main brands, Phonak, Unitron, Hansaton, Advanced Bionics, and several audiological care brands. Recently, the fourth core business consumer hearing was started with the brand Sennheiser. Here you can see some products of our comprehensive product portfolio. On the right side, cochlear implants, on the left side, hearing instruments, but also accessories. The consumer hearing devices are not yet on this picture. These are the main hearing instrument types, BTEs, RICs, and ITEs.
BTEs, Behind-The-Ears, they have the complete electronics, so battery, transceiver, transducer, signal processor in a housing behind the ear. They have an acoustic coupling with a sound tube and an ear mold, which goes inside the ear canal. Receiver in canal, as the name says, they have the receiver, which is the speaker in the ear canal. The rest, so the battery, microphones, and signal processor are still in housing behind the ear. Last but not least, ITEs, In-The-Ear devices. They have battery, transducer and signal processor in custom-made shells or housings in the ear. The current materials for custom products are acrylic, silicone, and titanium. Acrylic and titanium can be directly additively manufactured, so 3D printed. Silicone earpieces are produced with a cast process, where a cast is 3D printed and then filled with silicone.
We always keep our eyes open to investigate new or alternative materials and processes. Within such an investigative project, we are now studying a new acrylic material. This material consists, among other substances, of two sensitizing monomers. Therefore, we wanted to know whether the extracted concentrations are below acceptable limits to be compliant with biological safety according to ISO 10993. For one monomer, there's a relevant number of documented cases where it led to sensitization, acute contact dermatitis in medical devices. The EC3 is not known, but it is definitely below 5% because the lowest tested concentration in LLNA was 5% and resulted in a stimulation index of four, and a clear dose-response effect was observed. The other monomer has a calculated EC3 of 15.9%.
To get concentration-dependent data on sensitizing potential, which we can then use in a toxicological risk assessment, we wanted to apply the GARDskin Dose-Response method. These are the results of the GARDskin Dose-Response assay for the two monomers. As Andy explained before, a negative decision value means not sensitizing, a positive decision value means sensitizing. From these curves, we get the cDV0, the concentration where the decision value is zero. From this, EC3 values and NOAEL can be predicted. As expected, monomer one showed to be a strong sensitizer. Monomer two was as expected, so the EC3 was 22% versus the 16% found in literature. Now we can compare these values with extractable concentrations. If you want more information on the results themselves, they were presented in a poster at SOT.
I want to finish the presentation with scenarios for which in vitro sensitization methods can, and I think also should be used for medical devices. This is on one side, on R&D stage for the development of novel materials and proto-products, screening of material candidates, comparing different processing parameters, and also as pretests as a risk reduction before going to in vivo models. Also, as for generation of additional data in certain low-risk situations. For example, if we have minor changes or deviations in the material composition or processing, for the assessment of external influences which might occur over the product lifetime, assessment of materials and components with unlikely patient contact, or generally in situations where in the past only a cytotoxicity assay might have been performed. This is an example for the assessment of external influences, the impact of UVC light.
As a background, hearing aids can be placed in a UVC box for drying and cleaning process with the goal to reduce humidity and bacteria. However, UVC with wavelength of 100 nm-280 nm are able to degrade polymers. Therefore, we identified the risk that, we might have a chemical change of the material surface of a hearing aid or a component which may directly contact the end user's skin. According to MDR, which clearly requires that the biological safety risks need to be addressed with a focus on the complete lifetime, we need to assess the impact of the UVC exposure on the biocompatibility. This is how it was done. We did the first line testing with in vitro sensitization assays, DPRA and KeratinoSens assays. There, the extraction is done in aqueous solution and acetonitrile or DMSO.
In case of a positive or inconclusive result, we repeated the above testing also using non-exposed material as a reference. We did chemical characterization, comparing pre- and post-UVC exposed materials. We also did a GARDskin Medical Device testing because here the extraction in olive oil is possible, which might be very important as acetonitrile and DMSO are quite aggressive solvents. With this, I want to thank you for your attention, and we are now happy to answer your questions.
Thank you very much, Karla, for a really nice presentation and for demonstrating how the GARDskin Dose-Response assay can also be used in practice within the industry. By that, I would like to conclude today's presentation and thank you very much for taking the time to listen. Of course, thanks also to our invited guest speakers for their excellent presentation. I would like to open up the floor for a discussion and Q&A session. Thank you very much.
Yes, thank you. This was the recorded part of this webinar, and we are now moving over to the Q&A, as Andy was saying. As you can see, we have the presenters ready for your questions. We can say hello to Karla, to Rose-Marie, and to Andy. Great. Great to have you with us. As a reminder, if you want to submit a question, please do so by using the questions chat box. You'll find it in the GoToWebinar panel, and I can see that we have questions already. We should get going. Let's do the first question. Since you have already tested UVCB and multi-constituents, do you think that the GARD assay could also work for testing extracts?
Good question. Should I take that one? Yeah, I start.
You can start.
Yes. Thanks. Yes, I mean, we have quite some experience as well to use the assays for testing different type of natural extracts. They have been extracted using different type of solvents. We have also tested a bit oil-based lubricants and more complex samples. In the end, it's always context dependent, right, if it works or not. I mean, we have some data demonstrating that it works quite well for the ones we have tested this far. If we're not sure, I mean, what we can do initially is to perform a pre-validation exercise where we just basically do a pre-screening in the GARD assay to see whether it would be compatible or not.
There's no need from the beginning to complete the entire assay. We can just have a look at the substance, see if we can get some cytotoxicity, evaluate appropriate solvents, and also check the solubility properties if that would be an issue.
I also like to add that we do test extracts for medical devices. Making extracts from materials that you cannot dissolve is also a possibility. We can also follow the protocols that is used for the medical devices in the ISO standard.
Great. Thank you so much for explaining all that, Andy and Rose-Marie. The next question, for the work with agrochemicals, did you see any trend for false positive/negative with formulation type?
Good question.
Yes. Mm-hmm.
Yeah. Not really related to formulation type. There was a mixture of different type of formulations. These were both solvent-based and aqueous-based formulations. There was a different type of herbicides, fungicides. There was a lot of different type of everything from emulsions to full soluble materials in this initial subset of 20 substances. We did not see any link to the specific type of formulation in terms of false positive or false negatives. What I think is really interesting is that, as I also tried to demonstrate in the presentation, was that for the false positive, which were four of the substances, they were all basically borderline positive borderline negative classifications in Local Lymph Node Assay.
They induced a dose-response behavior in the assay, but not sufficient for passing the EC3 value of three. There is definitely a dose-response behavior as well in the LLNA assay for these materials. Some of them did also contain sensitizers. All of the four false positive classifications also contained known sensitizers within the formulations. It can also be explained by differences in sensitivity between the animal models and the in vitro assays, where the in vitro assays may be a bit more sensitive to be able to actually pick up that sensitizing signal. Unfortunately, there is no human data available for most of the formulations.
It's really hard to see whether the animal data or the in vitro assays is the most reliable in this case.
Okay. Thank you, Andy. Let's move on. Let's do the third question. Here it is. Any explanation for the ethylenediamine results that we're seeing?
Sorry, can you repeat that again?
Yeah, maybe I'm twisting my tongue when I try to read. An explanation for the ethylenediamine, I think that's how you pronounce it, results that we're seeing?
Oh, yeah. Thank you. Ethylenediamines.
Thank you.
It has been consequently false negative classification in GARDskin. It is a known prohapten, meaning that it will require some type of enzymatic activation in order to act as a viable hapten. Normally, and based on the other data we have for pro and prehaptens. The cells appear to be fit for purpose to predict also prohaptens in general, with ethylenediamine actually being the only false negative among these groups. It is a bit surprising. We are not sure why, especially since the other prohaptens seem to indicate that the metabolic activity of the cells are actually sufficient for predicting also prohaptens.
I don't have a really good explanation because as far as I remember as well, we could test it according to the standard protocol. There were no problems with neither cytotoxicity or solubility.
Yeah. Thank you so much, Andy. We're gonna move over to a couple of questions on dose response. One of them may be for the medical device team here. How do you recommend the results from dose response to be used?
Maybe you can answer that question, Karla, and explain how you are using it, and then I can fill in a bit more.
Okay. Looks like we have a sound problem here. Let me see if we can get Karla on. The microphone seems not to be working. Maybe you want to just switch the microphone in the audio section in the GoToWebinar tool and see if you can get it going again.
I think I'm back.
Yes. Great.
Okay. Sorry for that. It was kind of frozen. The way we use these results is, as I explained on one slide in R&D stage. If we develop new processes, look into new materials, if we want to have further data on the concentration-dependent sensitivity. We also use it if we have low-risk application situations, if we still have the same material, but maybe change something in the process or if we assess external influences, as I brought the example for the UVC curing. It's mainly things like this where we use the dose-response and in general, the in vitro sensitization assays. I can also add that we got a lot of questions around this when we went to the SOT meeting in the U.S. in March.
A lot of the chemists there that are doing a lot of chemical characterization of their materials and devices were really interested to learn more about this because they saw an opportunity to actually have a better risk assessment of the chemicals that they find that might be skin sensitizers. It's well known within the toxicology world and for medical devices that you cannot use a TTC concept for the local effects like skin irritation and skin sensitization. I believe that this, the GARDskin Dose-Response assay can be a tool that can be used in those cases where you have maybe a hit in an in silico model that it might be a skin sensitizer and further look into that specific chemical in that case.
Thank you.
Can I also?
Yes. Oh.
Can I also add one thing as well? Because there was a lot of focus there on medical devices. As we saw, from Christopher's presentation as well, I think it's important to highlight as well that another potential application is, as Christopher also was demonstrating, is to be able to have this more continuous prediction of the sensitizing potency. In that case, it can be used, for example, as a point of departure for a risk assessment procedure. Also to inform on which concentration where you should actually start your confirmatory HRIPT testing as well. I think that would be the main application for the cosmetic and fragrance industry at least, in the future.
Yeah, I mean, and that's also where we accumulate most data for the moment and why it's so important to demonstrate, yeah, the correlation to available local in vitro assay data and also to human data, I would say.
Great. Thank you so much. I know that we are approaching 5:00 P.M. but we still have. I think we have time to do at least one more question. It's another one that came in on the dose response, and it's simple. Can the dose response assay be used also for UVCBs and mixtures?
Yeah, good question. We are working to accumulate data to demonstrate also that it would work very well for UVCBs of different type. I mean, UVCBs is really diverse. It could be basically anything. So where we have focused the most is just on natural extracts, other type of materials that should be included into cosmetic products. But also, we have demonstrated and generated some data also for oil-based lubricants and other more complex samples. But from a technical perspective, it works perfectly fine. Similar to the local in vitro assay, you can also define the concentration based on a weight-based measure. So you don't really need to have access to a molecular weight.
In terms of testing mixtures, we have tested quite a few as well. There was mixtures included in the IFF dataset as well. We have also, yeah, quite a large internal database where we have evaluated mixtures as well. From a technical perspective, it works perfectly fine. There could, of course, be other issues dependent on the specific properties of the mixtures.
Okay. Thank you very much, Andy. That was actually the last question of today. We have reached the end of this webinar. If you have any further questions, please direct them to the email addresses that you can see on the screen right in front of you right now. It's for Dr. Andy Forreryd, Dr. Christopher Choi, and Dr. Karla Lienau. Thank you so much for participating in today's webinar, and thank you to Dr. Karla Lienau, Dr. Christopher Choi, Dr. Andy Forreryd, and Dr. Rose-Marie Jenvert. For your information, the recorded version of this webcast will be available on our website and will also be sent out to you by mail within the next couple of days. Thank you so much for now, and I wish all of you a good rest of the day. Bye-bye.
Thank you. Bye-bye.
Thank you. Bye.
Thank you very much.
Bye-bye.