BioVariance - Pediatric Pharmacogenomics in Drug Discovery
5 likes1,110 views
The document discusses pediatric pharmacogenomics, emphasizing the importance of understanding genetics and external factors for effective drug therapy in children. It highlights the need for integrated knowledge in drug discovery and treatment to address the significant unmet medical needs in pediatric populations. The presentation also outlines a systematic approach to using genetic data and machine learning to optimize drug dosages for children based on their unique physiological characteristics.
BioVariance - Pediatric Pharmacogenomics in Drug Discovery
- 1. Pediatric Pharmacogenomics – What knowledge do we have and how can we use it in bioinformatics analyses for Drug Discovery? Josef Scheiber, PhD www.biovariance.com m4 Seminar April 25, 2013
- 2. BioVariance - Overview Two distinct offerings rooted in the same data: • “Medical value content”-as-a-Service for Healthcare • making sense of genomic & other data in context as service offering Bio-Variance BaVarians ;) • We aim for testable hypotheses that are well- supported by data from scientific databases and the literature
- 3. Overview • General Introduction – Genetics – Impact of age (children) – Further influence factors • Key message • prediction-based example
- 4. Significant unmet medical need 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% diseases Drugresponserate NSAIDS 80 % response rate Alzheimer 25 % response rate Several thousand diseases without known treatment
- 5. Disease understanding getting better and better 2010 1970 1960 1950 Disease of the Blood Leukemia Chronic Leukemia Acute Leukemia Preleukemia Lymphoma Indolent Lymphoma Aggressive Lymphoma Increasing understanding of underlying biology opens up new hypotheses 5 Year Survival ~ 0 % ~ 70% Example: Leukemia and Lymphoma
- 6. But still, mostly off-label use for children 3.9 1.4 3.4 0 1 2 3 4 5 6 7 Inlabel / Licensed Offlabel (Unlicensed) Adverse Drug Reactions of unlicensed / off-label prescription in pediatrics In hospital (UK) Outpatient clinics (FR) (Turner et al, Acta Paed, 1999) (Horen et al, Brit J Clin Pharm, 2002)
- 7. Integrated Knowledge is key for good interpretations Genetics Age External influence factors
- 8. Integrated Knowledge is key for good interpretations Genetics Age External influence factors
- 9. There are ~7 billion human genomes and each responds differently to drugs Medically actionable annotations are key, particularly in the area of Pharmacogenetics
- 10. Knowledge is key to enable decisions – both for Drug Discovery and Treatments ?
- 11. Impact of genetics on Drug Action The human genome contains roughly 3 billion nucleotides and the genomes of any 2 individuals vary in 3 million of them A significant likelihood that individuals respond differently to the same medicine This is rooted in differences for drug absorption, distribution, metabolism and excretion
- 12. Examples: Genes impacting Drugs Biotransformation: Phase I (Oxidation, Reduction, Hydrolysis, Hydration, Dethioacetylation, Isomerization) Phase II (Glucuronidation, Sulfation, Methylation, Acetylation, Amino Acid Conjugation, Gluthathione Conjugation, Fatty acid conjugation) Gene Drug Bcr/abl or 9:22 translocation Imatinib HER2-neu Trastuzumab EGFR mutations Gefitinib Thiopurine S-methyltransferase Mercaptopurine, Azathioprine UGT1A1 Irinotecan CYP2D9/VKORC1 Warfarin HLA-B*5701 Abacavir HLA-B*1502 Carbamazepine CYP2C19 Clopidrogel …. Many more
- 13. Ultimately: An individual profile
- 14. Example: Carbamazepine/Steven Johnsons Syndrome Courtesy: Dr. Thomas Habif dermnet.com A single mutation can have massive impact Difference in European and Korean populations
- 15. On a bigger scale – similar impact • Example: Treatment of ALL (acute lymphoblastic leucemia) – Patients with ALL who have 1 wild-type allele and intermediate TPMT activity tend to have a better response to 6MP (Mercaptopurine) therapy than patients with 2 wild-type alleles and full activity – Pharmacogenetic polymorphisms of several additional genes also have the potential to influence successful treatment of ALL – 20% of patient with ALL who do not respond to chemotherapy represent an additional challenge for pharmacogenomic research
- 16. Understand the link between Types of Genetic Variation and phenotype • Single Nucleotide Aberrations – Single Nucleotide Polymorphisms (SNPs) – Single Nucleotide Variations (SNVs) • Short Insertions or Deletions (indels) • Larger Structural Variations (SVs)
- 17. Classes of structural variation Alkan, C. et al. Genome structural variation discovery and genotyping. Nature Reviews Genetics 12, 363-376 (2011).
- 18. Raw Data Analysis Image Processing and base calling Whole Genome Mapping Alignment to reference genome Variant Calling Detection of genetic variation (SNP, CNV etc.) Annotation Linking variants to biological information BioVariance focus Quick detour: Basic NGS workflow
- 19. Integrated Knowledge is key for good interpretations Genetics Age External influence factors
- 20. Age-related influence factors on Drug Therapy Physiologic Factors that influence the Oral absorption of Medications PARAMETER Neonate Infant Child Gastric Acid secretion Reduced Normal Normal Gastric Acid Emptying Time Decreased Increased Increased Intestinal Motility Reduced Normal Normal Biliary Function Reduced Normal Normal Microbial Flora Acquiring Adult Pattern Adult pattern
- 21. Premature Neonate Neonate Infant Child Adolescent Absorption Gastric acidity Decreased Decreased Decreased Equal Equal Gastric emptying time Decreased Decreased Equal Equal Equal GI motility Decreased Decreased Decreased Equal Equal Pancreatic enzyme activity Significantly decreased Decreased Decreased Equal Equal GI surface area Increased Increased Increased Increased Equal Skin permeability Significantly increased Increased Equal Equal Equal Distribution Body composition Equal Blood-brain barrier Decreased Decreased Equal Equal Equal Plasma proteins Significantly decreased Decreased Equal Equal Equal Metabolism Liver Decreased Decreased Decreased Equal/Increased Equal Elimination Renal blood flow Decreased Decreased Decreased Equal Equal Glomerular filtration Decreased Decreased Decreased Equal Equal Tubular function Decreased Decreased Decreased Equal Equal
- 22. Oral Drug Absorption in the Neonate vs Older Children and Adults – no direct heigt/weight etc. correlation Drug Oral Absorption Acetaminophen Decreased Ampicillin Increased Diazepam Normal Digoxin Normal Penicillin G Increased Phenobarbital Decreased Phenytoin Decreased Sulfonamides Normal
- 23. Expression patterns different between age groups - There are many literature examples for many relevant indications way too many to detail them here - However, nobody has yet attempted to integrate genetic and pediatric information on a large scale, this is what we are doing now
- 24. Integrated Data is Key for good interpretations Genetics Age External influence factors
- 25. Drug-Drug Interactions • When 2 or more drugs are administered to the same patient, the pharmacokinetic and pharmacodynamic properties of each agent may be modified by their interaction. – Acetaminophen + alcohol = Increase hepatotoxicity – Antacid + Iron = decrease absorption – Digoxin + Cimetidine = Increase Digoxin toxicity
- 26. Tobacco drug interactions Food drug interactions
- 27. • A full integration of genetic and additional data is currently missing, we are addressing this in a collaboration project • If you want to develop a drug for a stratified population, you want to understand the interplay of all these factors in detail ∑
- 28. An example
- 29. Incorporating pediatric data into predictive approaches Starting point: Netdosis (www.netdosis.de) has collected a dataset of pediatric on- and off-label use for many different medications - Data is very well structured and therefore amenable for machine- learning approaches - Drug names linked to machine-readable description of chemical structures
- 30. The idea • Link chemical substructures to their influence on pediatric dosage • Use this information to predict dosage levels for not yet tested drugs and drug candidates
- 31. Simplified workflow (1) Predict dosage levels at different age ranges (2) Investigation of dosage-related information for hypothesis development
- 32. Data input • Your compounds • Netdosis database • Your internal data incorporated where applicable • Specifically curated scientific papers around particular usages (especially if some interesting facts turn up in first run)
- 33. Computational description of molecules • Descriptor selection heavily impacts outcome of analyses • Depending on your main objectives different technologies are the best fit, we will discuss this in detail with you 0 1 0 0 2 0 0 0 1 0
- 34. Statistical modeling • Activity is either in categories (age range) or more granular depending on your needs • Plenty of positive results with naïve Bayes, therefore method of choice • Other technologies depending on data/on request • Strict model validation
- 35. n compounds defined activity 1 3 n model predict 2 3 nrepeat3xtimes 2 3 n Internal measure for model quality R2 CV-50% Predictionmodel training data set test data set Model Validation - Example
- 36. 1 3 n model predict 2 3 nrepeat3xmal 2 3 n Internal measure for model quality R2 CV-50% PredictionModel External measure for model quality R2 Test,Avg repeat atleaste100x Model Validation - Example
- 37. Prediction results • Based on model sets for each dosage level and age range, there are 100 prediction results for each dosage level/age range • These are further analyzed, usually median predictive value taken for prediction and ranking • Result: A ranked list with associated probabilities for each dosage level/age range T1 T2 T3 … …
- 38. What does the result mean? • Targets need to be annotated with phenotypic outcome – i.e. what does it mean that the compound is hitting this target? • Do we have opportunities ( repurposing) or liabilities ( side effects) or both? • How do different compounds compare? • What predictions should be confirmed by testing?
- 39. Possible extensions Diving into chemical biology • Map into pathways • Retrieve marketed drugs and clinical candidates that act in these pathways
- 40. Dealing with a very complex environment – i.e. many opportunities DNA RNA Protein Interactions Clinical parameters Treatment History Tissue anatomy Surgical History Epigenetic Profiles from many patients at different timeponits Target Off-targets Metabolites Additional indications Unspecific effects Similar drugs Adapted from: J. Scheiber; How can we enable drug discovery informatics for personalized healthcare? Expert Opinion on Drug Discovery, 1-6; 2/2011
- 41. Outlook The right drug for the right patient at the right time & right dose is only possible if you have the right knowledge within the right context right in place We will further work on this!
- 42. Thank you for your attention! josef.scheiber@biovariance.com Phone: +49 – 89 – 189 6582 – 80 Garmischer Str. 4/V 80339 Munich / Germany