Senzagen Ab · OMXSTO:SENZA
Senzagen Ab — Technical Presentation (April 2019)
61 slides from Senzagen Ab's technical presentation, April 2019
Slide 1: Genomics and Machine Learning – All you need to know about the GARD technology
Genomics and Machine Learning – All you need to know about the GARD™ technology for in vitro sensitization testing Copyright Protected and all rights reserved. This is the property of SenzaGen and must not be reproduced in part or full without SenzaGen’s written permission.
Slide 2: About SenzaGen
About SenzaGen GARD® • Founded in 2010 • Business model: global Our lead product, GARD® stands • Spin-out from Lund University industrial and CRO partnerships for Genomic Allergen Rapid after over 10 years research • Own laboratory, continuous Detection and is a state of the art • Highly multidiscip
Slide 3: Predictive toxicology: a scientific field in change
Predictive toxicology: a scientific field in change Predictive Simple workflow In vitro performance In vitro GARD In vivo Copyright Protected and all rights reserved. This is the property of SenzaGen and must not be reproduced in part or full without SenzaGen’s written permission.
Slide 4: High informational content allows for building scientific models closer to reality
High informational content allows for building scientific models which are closer to reality Increased number of biomarkers Copyright Protected and all rights reserved. This is the property of SenzaGen and must not be reproduced in part or full without SenzaGen’s written permission.
Slide 5: High informational content allows for building scientific models closer to reality
High informational content allows for building scientific models which are closer to reality S Sensitizer NS Non-Sensitizer Y S S S S NS NS Increased number S NS NS Sof biomarkers NS S X Biomarker 1 Copyright Protected and all rights reserved. This is the property of SenzaGen and must not be reprodu
Slide 6: High informational content allows for building scientific models closer to reality
High informational content allows for building scientific models which are closer to reality S Sensitizer NS Non-Sensitizer S S S Z S S S Y Biomarker 3 S NS NS Increased NS number NS of biomarkers NS X Biomarker 1 Copyright Protected and all rights reserved. This is the property of SenzaGen and must
Slide 7: The GARD platform – a modular framework for testing toxicological endpoints
The GARD platform – a modular framework for testing of various toxicological endpoints Prediction model Genomic biomarker signature Training data Biological system Copyright Protected and all rights reserved. This is the property of SenzaGen and must not be reproduced in part or full without SenzaGe
Slide 8: The GARD platform – Currently available assays
The GARD platform – Currently available assays GARD™skin (200 genes) Skin sensitization testing GARD™potency (51 genes) Skin sensitization potency according to GHS/CLP GARD™air (28 genes) Respiratory sensitization testing Copyright Protected and all rights reserved. This is the property of SenzaGen
Slide 9: The GARD platform – One testing platform for all your answers
The GARD platform – One testing platform for all your answers Copyright Protected and all rights reserved. This is the property of SenzaGen and must not be reproduced in part or full without SenzaGen’s written permission.
Slide 10: The GARD platform – biological system
The GARD platform – biological system Biological system Copyright Protected and all rights reserved. This is the property of SenzaGen and must not be reproduced in part or full without SenzaGen’s written permission.
Slide 11: The GARD platform – biological system
The GARD platform – biological system Allergic Contact Dermatitis Dendritic cells (DCs) 1. Hapten-protein complex 2. Inflammatory mediators Local lymph node 4. T- cell proliferation 3. DC activation Allergen specific T-cells Copyright Protected and all rights reserved. This is the property of SenzaG
Slide 12: The GARD platform – biological system
The GARD platform – biological system Keratinocytes Dendritic cells T-cells Key Key Event 3 KeyEvent 1 1 Event Key Key Event Event 21 Key Event 4 Adverse Outcome Hapten formation Inflammatory >DC activation Activation/ Allergic Contact Covalent binding mediators >DC maturation proliferation of Derma
Slide 13: The GARD platform – biological system
The GARD platform – biological system Dendritic cells (DCs) Respiratory sensitization Skin sensitization T-cells: CD4+ Th2 T-cells: CD4+ Th1 & CD8+ Tc Humoral Immunity Cellular Immunity naive T-cells Copyright Protected and all rights reserved. This is the property of SenzaGen and must not be reprod
Slide 14: What are SenzaCells?
The GARD platform – biological system What are SenzaCells? SenzaCell’s: A human DC-like cell line Antigen presentation: MHC I, MHC II Cd1d. Potent activators of T-cells: CD80, CD86 T-cell polarization: Contains the functional elements for Th1 & Th2 polarization. Copyright Protected and all r
Slide 15: The GARD platform – biological system
The GARD platform – biological system DNA Transcriptomics mRNA Proteins Recognition of foreign Activation of self- Cellular stress Communication with substances. defense mechanisms. responses. cells of the immune system. Copyright Protected and all rights reserved. This is the property of SenzaGen a
Slide 16: Assay Development: Hypothesis
The GARD platform – biological system Assay Development: Hypothesis There should exist genes and pathways in DCs that are differentially expressed depending on the stimuli Such genes could be used as predictive tools Skin Sensitizer Cellular responses Respiratory Sensitizer Non-Sensitizer Copyri
Slide 17: The GARD platform – training data
The GARD platform – training data Training data Biological system Copyright Protected and all rights reserved. This is the property of SenzaGen and must not be reproduced in part or full without SenzaGen’s written permission.
Slide 18: Assay Development: Defining a training data set
The GARD platform – training data Assay Development: Defining a Training data set to test the Hypothesis. Training dataset: A set of well-characterized chemical compounds with known expected outcome. X number of Sensitizers X number of Non-sensitizers Ideally, the training set should cover a large c
Slide 19: The GARD platform – training data
The GARD platform – training data 1 Chemical stimulations 2 RNA extraction Prepare chemicals for Stimulate cells for 24 h Extract RNA from cells cellular exposure 3 Global transcriptomic profiling Gene expression Measure complete transcriptomes of cells using microarrays > 29.000 genes/sample Quanti
Slide 20: The predictive Genomic Biomarker Signature
The GARD platform – the predictive Genomic Biomarker Signature Genomic biomarker signature Training data Biological system Copyright Protected and all rights reserved. This is the property of SenzaGen and must not be reproduced in part or full without SenzaGen’s written permission.
Slide 21: The predictive Genomic Biomarker Signature
The GARD platform – the predictive Genomic Biomarker Signature Genomic biomarker signature Gene 1 Gene 2 ………… Data-driven biomarker identification Data analysis is used to identify genes that are differentially regulated depending on whether cells are stimulated with Sensitizers or Non- sensitizers.
Slide 22: The predictive Genomic Biomarker Signature
The GARD platform – the predictive Genomic Biomarker Signature Genomic biomarker signature # genes: > 29.000 genes # genes: Custom – up to 800 genes Assay time: 4 days Assay time: 22 h, 15 min hands-on Throughput: 24 samples/batch Throughput: 384 samples/day Input: cDNA Input: Total RNA All genes ar
Slide 23: The predictive Genomic Biomarker Signature
The GARD platform – the predictive Genomic Biomarker Signature NanoString Affymetrix Non-Sensitizers Sensitizers Non-Sensitizers Sensitizers Spearman correlations for each stimulation between 85-90% Gene expression measurement could be reproduced on the independent platform Copyright Protected and a
Slide 24: The GARD platform – Prediction model
The GARD platform – Prediction model Prediction model Genomic biomarker signature Training data Biological system Copyright Protected and all rights reserved. This is the property of SenzaGen and must not be reproduced in part or full without SenzaGen’s written permission.
Slide 25: Assay Development: Hypothesis
The GARD platform – Prediction model Assay Development: Hypothesis There should exist genes and pathways in DCs that are differentially expressed depending on the stimuli Such genes could be used as predictive tools Skin Sensitizer Cellular responses Respiratory Sensitizer Non-Sensitizer Copyrig
Slide 26: Use modern technology for pattern recognition
The GARD platform – use modern technology for pattern recognition 1 Machine learning is a computer What is algorithm designed to learn from it? previous experience. 2 By observing data from a set of How objects (”training data”) with known does it expected outcomes (”labels”) . work? Training 3 The
Slide 27: Use modern technology for pattern recognition
The GARD platform – use modern technology for pattern recognition 2 By observing data from a set of How objects (”training data”) with known does it expected outcomes (”labels”) . work? Gene 1 The training data: Gene 2 ………… Gene expression profiles from a set of well-characterized sensitizers and no
Slide 28: Use modern technology for pattern recognition
The GARD platform – use modern technology for pattern recognition Training 3 The machine learning algorithm learns. The machine learning algorithm assigns a label to the new data. 4 Class- ification Fixed Formula ? OR NS NS S NS S NS S S S Sensitizer NS Non-Sensitizer The compound is a Fixed Formula
Slide 29: Support Vector Machines
The GARD platform – use modern technology for pattern recognition Support Vector Machines (SVM) Decision value > 0 = Sensitizer S Sensitizer (Training data) Decision value < 0 = Non sensitizer NS Non-Sensitizer (Training data) ? Classified as a Sensitizer ? Unknown (Test data) Sensitizers Non-sensit
Slide 30: Hazard prediction
Hazard prediction Prediction model Genomic biomarker signature Training data Biological system Test substance Copyright Protected and all rights reserved. This is the property of SenzaGen and must not be reproduced in part or full without SenzaGen’s written permission.
Slide 31: How to GARD your product in steps
How to GARD your product - in 6 Steps Copyright Protected and all rights reserved. This is the property of SenzaGen and must not be reproduced in part or full without SenzaGen’s written permission.
Slide 32: The GARD platform – Currently available assays
The GARD platform – Currently Available assays GARD™skin (200 genes) Skin sensitization testing GARD™potency (28 genes) Skin sensitization potency according to GHS/CLP GARD™air (51 genes) Respiratory sensitization testing Copyright Protected and all rights reserved. This is the property of SenzaGen
Slide 33: GARDskin – Assay development: the training set
GARDskin – a highly accurate assay for skin sensitization testing Assay development – The training set SenzaCells were stimulated with a reference panel of 20 Skin sensitizers 20 Non-sensitizers =29,000 genes/sample Training data Biological system Copyright Protected and all rights reserved. This is
Slide 34: GARDskin – Training data set and GARD prediction signature
GARDskin – a highly accurate assay for skin sensitization testing Training Data set GARD prediction signature (GPS) 20 Skin sensitizers 20 Non-sensitizers Prediction model Genomic biomarker signature Non-Sensitizers Sensitizers Training data Biological system 200 genes Copyright Protected and all ri
Slide 35: GARDskin – Full transparency
GARDskin – a highly accurate assay for skin sensitization testing Full transparency: All assay development steps & the complete list of genes have been published. Copyright Protected and all rights reserved. This is the property of SenzaGen and must not be reproduced in part or full without SenzaGen
Slide 36: GARDskin – Accuracy, sensitivity and specificity
GARDskin – a highly accurate assay for skin sensitization testing Data set Sensitivity Specificity Accuracy Reference GARD in-house validation 89% (17/19) 86% (6/7) 88% (23/26) Johansson, 2014 Technology transfer and method 94% (16/17) 83% (10/12) 90% (26/29) Forreryd, 2016 optimization CE Evaluatio
Slide 37: GARDskin validation – Validation Phase I
GARDskin validation – Validation Phase I Training & Transfer Phase Validation Phase I N = 28 BRT (US) blinded compounds Lead Laboratory SenzaGen N = 28 blinded (SE) compounds Eurofins N = 28 blinded (EU) compounds GARDskin Copyright Protected and all rights reserved. This is the property of SenzaGen
Slide 38: GARDskin validation – Validation Phase I results
GARDskin validation – Validation Phase I Experiment 1 Sensitizers Non-sensitizers Experiment 2 20 Experiment 3 15 10 5 Burleson Research GARD Decision Value 0 Within Laboratory Reproducibility: Technologies -5 SenzaGen 82% 20 Burleson 83% Accuracy 94% 15 Eurofins 89% Sensitivity 93% 10 5 Between Lab
Slide 39: GARDskin – Added values of ITS: True or False?
GARDskin – Added values of ITS: True or False? DPRA ARE-NRF2 h-CLAT 2 out of 3 GARD (TG 442C) (TG 442D) (TG 442E) ITS (TGP 4.106) Accuracy 80% 83% 77% 83% 88% Sensitivity 78% 84% 80% 84% 90% Specificity 83% 78% 67% 78% 83% Roberts, D.W. John Moore University, Liverpool. Regul Toxicol Pharmacol, 2018
Slide 40: GARDskin – high informational content & broad mechanistic coverage
GARDskin – high informational content & broad mechanistic coverage GARD prediction signature (GPS) Recognition of foreign substances e.g. TLRs, RXR, AHR Immunological self-defence mechanisms e.g. CD80, CD86 Cellular stress responses e.g. NRF2-pathway Communication Non-Sensitizers Sensitizers e.g. ch
Slide 41: GARDskin – high informational content & broad mechanistic coverage
GARDskin – high informational content & broad mechanistic coverage Keratinocytes Dendritic cells T-cells Key Event 1 Key Event 2 Key Event 3 Key Event 4 >Pro-inflammatory >Antigen recognition Covalent binding to signalling >DC activation Activation/ proteins >Cytoprotective >DC maturation proliferat
Slide 42: GARDskin – Broad applicability domain
GARDskin – Broad applicability domain GARDskin, 100 applicable for: Mean accuracy (89%) 80 • Agrochemicals Active Pharmaceutical Ingredients • Cosmetics 37 Disinfectants • Industrial chemicals Dyes • UVCBs • Natural extracts Accuracy (%) Flavoring agents/Frangrances • Screening of pharma- Intermedia
Slide 43: GARDskin – what about difficult to test substances?
GARDskin – what about ”difficult to test” substances? Why substances can be difficult to test: Pre/Pro haptens: Activation may be needed to create the allergen. SenzaCells: Aldehyde dehydrogenases (ALDH) Cytochrome p-450 (CYP) N-acetyltransferase 1 (NAT-1) Solubility: High logP value or other pr
Slide 44: GARDskin – difficult to test substances comparison
GARDskin – what about ”difficult to test” substances? Compound Pre/pro High DPRA ARE-NRF2 h-CLAT GARD hapten logP (TG442C) (TG 442D) (TG442E) (TGP 4.106) 2-Aminophenol YES - S S S S1 2-nitro-1,4-Phenylendiamine YES - S S S S1 Aniline YES - NS NS S S1 Cinnamic alcohol YES - S S S S1 Ethylene diamine
Slide 45: The GARD platform – Available assays
The GARD platform – Available assays GARD™skin (200 genes) Skin sensitization testing GARD™potency (28 genes) Skin sensitization potency according to GHS/CLP GARD™air (51 genes) Respiratory sensitization testing Copyright Protected and all rights reserved. This is the property of SenzaGen and must n
Slide 46: GARDpotency – subclassification of weak and strong skin sensitizers
GARDpotency – subclassification of weak and strong skin sensitizers Training Data set GARDpotency prediction signature -2 22 Class 1A (Strong Sensitizers) 29 Class 1B (weak-sensitizers) 0 2 Prediction model 28 Genomic biomarker signature genes Training data Biological system Weak Strong Copyright Pr
Slide 47: GARDpotency – subclassification of weak and strong skin sensitizers
GARDpotency – subclassification of weak and strong skin sensitizers GARDskin GARDpotency 1A Sensitizer Test 1B substance Non- Sensitizer Copyright Protected and all rights reserved. This is the property of SenzaGen and must not be reproduced in part or full without SenzaGen’s written permission.
Slide 48: GARDpotency validation – Validation Phase II
GARDpotency validation – Validation Phase II Training & Transfer Phase Validation Phase I Validation Phase II N = 28 BRT (US) blinded compounds Lead Laboratory SenzaGen N = 28 blinded N = 30 (SE) compounds blinded compounds Eurofins N = 28 blinded (EU) compounds GARDskin GARDpotency Copyright Protec
Slide 49: GARDpotency validation – Validation Phase II results
GARDpotency validation – Validation Phase II Experiment 1 Experiment 2 Weak sensitizers (1B) Strong sensitizers (1A) Experiment 3 Burleson Research GARD Decision Value Technologies Accuracy 82% Correct 1A 89% Eurofins Correct 1B 73% SenzaGen Copyright Protected and all rights reserved. This is the p
Slide 50: GARDskin & GARDpotency – REACH registration
GARDskin & GARDpotency – REACH registration “The REACH Regulation allows the use of non-adopted in vitro methods in case they meet the EURL ECVAM criteria for entering pre-validation. For the GARD assay this criteria is met, as it is currently being validated.” “The current REACH information require
Slide 51: The GARD platform – Available assays
The GARD platform – Available assays GARD™skin (200 genes) Skin sensitization testing GARD™potency (28 genes) Skin sensitization potency according to GHS/CLP GARD™air (51 genes) Respiratory sensitization testing Copyright Protected and all rights reserved. This is the property of SenzaGen and must n
Slide 52: GARDair – Identification of chemical respiratory sensitizers
GARDair - Identification of chemical respiratory sensitizers Anhydrides Respiratory sensitizers are chemical compounds capable of inducing an allergic hypersensitivity reactions in the airways. Isocyanates Respiratory sensitizers often elicit a positive response in in vivo (and in vitro) models
Slide 53: GARDair – EU Horizon 2020 funding
GARDair - Identification of chemical respiratory sensitizers EU Commission – Flagship Product Health : “GARDair – The first predictive in vitro assay for the identification of respiratory sensitizers. “ €2.4 M in funding HORIZON 2020 The EU Framework Programme for Research and Innovation This projec
Slide 54: GARDair – Training data set and prediction signature
GARDair - Identification of chemical respiratory sensitizers Training Data set GARD respiratory prediction signature 10 respiratory sensitizers 20 Non-respiratory sensitizers (incl. skin sensitizers) Prediction model Genomic biomarker signature 51 genes Training data Biological system Non-Respirator
Slide 55: GARDair – coverage of mechanistically relevant pathways
GARDair – coverage of mechanistically relevant pathways Keratinocytes Dendritic cells T - cells Key KeyEvent 1 1 Event Key Key Event Event 21 Key Event 3 Key Event Key Event 44 Adverse Outcome Hapten formation Inflammatory >DC activation Activation/ Occupational > covalent binding mediators >DC matu
Slide 56: GARDair – in-house validation & performance
GARDair - in-house validation & Performance Skin sensitizers? Accuracy 90% Sensitivity 100% Specificity 88% Test substance with mean GARD decision value from biological triplicate samples > 0 = Respiratory Sensitizer Copyright Protected and all rights reserved. This is the property of SenzaGen and m
Slide 57: GARDair – in-house validation & performance
GARDair - in-house validation & Performance Accuracy 90% Sensitivity 100% Specificity 88% Copyright Protected and all rights reserved. This is the property of SenzaGen and must not be reproduced in part or full without SenzaGen’s written permission.
Slide 58: GARDair – ring trial initiated on external laboratories
GARDair - ring trial initiated on external laboratories Training & Transfer Phase Ring trial Burleson N = 30 blinded (US) compounds Lead Laboratory SenzaGen N = 30 blinded (SE) compounds Eurofins N = 30 blinded (EU) compounds Copyright Protected and all rights reserved. This is the property of Senza
Slide 59: How can I get my substances tested?
The GARD platform – how can I get my substances tested? Contact us: We help to design the testing strategy for your specific substances. Select assay(s): GARDskin, GARDpotency, GARDair. Test Substance Questionnaire: Do you have information on vehicles? If not, we evaluate it for you. Turnaround time
Slide 60: Where can I get my testing done?
The GARD platform – where can I get my testing done? Copyright Protected and all rights reserved. This is the property of SenzaGen and must not be reproduced in part or full without SenzaGen’s written permission.
Slide 61: Information and contacts
Information and contacts www.senzagen.com Andy Forreryd, PhD Joshua Schmidt, PhD Key Account Manager & Scientific Liaison Business Development Director Americas andy.forreryd@senzagen.com Joshua.schmidt@senzagen.com +46 734331777 +1 651.440.5691 Copyright Protected and all rights reserved. This is t