The Personalization of Medicine Data per patient W hat can Precision - PowerPoint PPT Presentation

Ro w Analytics transforming the delivery of health Steve Gardner, CEO E: steve@rowanalytics.com T: +44 1865 575170 / +44 7799 671644 W: rowanalytics.com

The Personalization of Medicine Data per patient

W hat can Precision Medicine Deliver? US Health costs/yr per person 1 US population aged 65+ 2 88.5M £30K Patients with long-term conditions on 3+ drugs cost 5x more and have poorer outcomes 36.7M £4K Age 2005 2025 2050 20 50 80 Precision medicine can deliver: 48% 75% 40% fewer ER visits 3,6 better patient fewer readmissions 3,6 reported outcomes 4 £225B+ /yr compliance related savings to 1 CMS.gov (2014) 2 Kaiser Family Foundation (2015) The Rising Cost of Living Longer US healthcare 5,6,7 3 Clinical impact of pharmacogenetic profiling (2017) Elliott LS et al. PLoS ONE 12(2) 4 Walsh & Cussen, Ir Med J, 2010 103(8):236-8 5 Viswanathan et al Interventions to Improve Adherence (2012) Annals of Internal Med 6 Sultana, J., Cutroneo, P., Trifioro, G. (2013). J. Pharmacology & Pharmacotherapeutics 7 Brixner, D., et al Effect of pharmacogenetic profiling (2015). J Med Economics 19, 3

Healthcare Industry Megatrends • Payment on results – more accurate diagnosis & Outcomes evidence of cost-effective patient benefit Precision • Giving the right treatments to the right patient at Medicine the right time, every time Patient • Helping patients & carers be better informed and Engagement actively change behavior to manage their health

Whole Genome Analysis

Precision Medicine Use Case SEQUENCE IDENTIFY ANALYZE DECIDE patient mutations metabolic best drug genome (SNPs) function prescription World’s fastest and Powerful and Most personalized most scalable efficient deep clinical decision genome association semantic learning support & digital studies & search tools health platform

‘Traditional’ Biomarker Discovery / GWAS • Correlate genetic markers with disease/treatment outcomes Source: MMG 233 2014 Genetics & Genomics Wiki

The Biological Challenge

The Computational Challenge • Current fastest supercomputer does 3 x 10 16 operations/sec • At n = 6 would take 3.1 trillion years… • At n = 10 would take 3.6 x 10 38 seconds n! 3 r / r! (n − r)!

Bipolar Study Findings 607 Layer # Networks Features % Cases % Controls (# SNPs in (cumulative) (cumulative) combination) Bipolar patients 1 - - - 1,355 2 - - - 3 3 Rare variant homozygote 0 Controls 4 1 Rare variant homozygote 26% 0 803 (158/607) 5 - SNPs per person 6 - 1.7x10 28 7 - 8 - possible combinations 9 - 10 1 Common variant 44% 0 homozygote/ heterozygote (222/607) The n-SNP networks are genetically distinct and have been clinically validated

Genomics Data Representation SNP Genotype: 0 = homozygous ‘normal/major allele’ SNP 247 0 247 0 247 0 247 0 1 = heterozygous 2 = homozygous ‘variant/minor allele’ Genotype SNP Index 247 = rs12345678 Case #27 … 1 0 2 0 3 1 4 0 5 2 6 1 n 0 Genotype A C B States 218 1 775 1 Layer 3 247 0 3 1 5 2 3 1 3 1 5 2 842 1 (SNP Genotypes) (Case Indices) [3, 6, 8, 9, 14, 56] [2, 7, 9, 10, 23] [1, 4, 56, 99, 113] where nCases => minCases (e.g. 5 above) and nControls <= maxControls (e.g. 0)

Synomics Example – Breast Cancer Study 14,777 200K People with SNPs per person BRCA1/2 mutations • All participants have BRCA 1 and/or BRCA 2 mutations 3,850 affected by breast cancer (cases) • 10,927 non-affected (controls) • • Seeking combinations of multiple SNPs associated with: disease risk • disease protective effect • therapy response •

Synomics – Transforming Genomic Medicine 14,777 200K People with SNPs per person BRCA1/2 mutations synomics Current GWAS (on single IBM Minsky with (1,000 node supercomputer) Comb. 4x Nvidia P100) 10 11 12 mins 2 SNPs 6-8 months 10 32 6 hours 6 SNPs - 17 SNPs 10 84 - 6 days

Synomics - Breast Cancer (BRCA1)

BRCA2 affected / non-affected 1,576 6,402 200K Affected (cases) Non-Affected (controls) SNPs per person False Discovery SNP Cases Penetrance Rate (%)* Genotypes (%) 20 2,113 799 50.7 10 1,320 627 39.8 5 868 513 32.6 1 142 221 14.0 * Using Benjamini-Hochberg correction for multiple testing

Synomics Example – Breast Cancer Study Key Findings: Found co-occurring sub-clusters of 3, 4, 5, or 6 SNP variants • co-occurring in later layers of analysis (8 SNPs+) • SNPs associated with same pathways show disease functional units • opportunity to identify combinatorial therapies Detected 17 SNP networks in up to 103 cases and 0 controls Very high (>25%) penetrance for good clinical relevance Identified disease protective & disease risk associated factors • BRCA1/2 status may suggest risk, but other variants in combination confer an overall greater protective effect Currently analysing phenotypic and clinical features

Genomics Data Representation Merged [A, B, C, D] 3 1 [B, E] 218 1 [A, B, C, D] merged cases 3 1 Networks [A, C] 5 2 [1, 2, 3, 4, 6, 7, 8, 842 1 [C, F] [A, C] (based on 5 2 9, 10, 14, 23, 56, [A, D] 247 0 [B, F] 775 1 sharing at 99, 113, 246, 299, least x% SNP [A, D] …] [D, E] [E, F] 247 0 264 1 511 2 genotypes) Merged A Networks 3 1 3 1 , 5 2 higher nCases with lower nSGs and high densities is better, i.e. a small 3 1 number of highly interconnected C B SNP genotypes 3 1 , 247 0 3 1 3 1 D

Real World Personalisation Challenges PERSONALISED MEDICINE 101 • Clinical status • Phenotype • Co-morbidities • Co-prescriptions • Lifestyle & environment More accurate assessment of diagnosis and response to treatment. • (Clonal) heterogeneity Molecular profiling is used to determine the appropriate therapy. • Polygenic disease aetiology Adapted from: sarahcannon.com

• 15,000 MND patients / 7,500 controls • 40% whole genomes sequenced • 2 petabytes • Multi-factor late-onset disease • Only 5-10% genetically determined heritability • 6 independent factors required to trigger disease • Imaging, epigenetics, lifestyle, diet, environment clinical history, co-morbidities

Biological Interpretation

Biological Annotation • Query: rs3734805 rs9383935 rs9383589 c6_pos151989450 rs4648881 rs9383936 CD14+CD16- monocyte CD8+/ab T fetal thymus naive B cell • Full context of all cell types in which epigenetic activation occurs • Literature search (keywords) gave no relevant results (Google/PubMed) • Deep semantic search identified 36 relevant papers including: “ In vivo modulation of the distribution of thymocyte subsets: effects of estrogen … ” • Further queries identified a study where female infants with enlarged thymus treated with X-rays were observed to have higher incidence of breast cancer 36 years later • Suggested novel disease sensitization mechanism

spot .my – deep semantic matching • matches all of the words to find the best, most relevant hits • more words = better context = better hits “Transient involution of the maternal thymus in mice is known to use keywords occur during pregnancy. Although estrogen crosses the placenta, fetal or whole thymus gland enlarges with advancing gestational age. It is not paragraphs to known if fetal thymocytes are resistant to estrogen or if there are search other factors that prevent estrogen from exerting an effect on the development of fetal thymocytes. Therefore we studied the effect of estrogen on isolated fetal thymic glands in vitro. All CD4 and CD8 defined T cell subsets were reduced with a disproportionate loss of CD4+ single positive (SP), CD8+ SP: CD4+CD8+ double positive (DP) cells.”

spot.my GPU enabled semantic search documents manual content doc queue worker indexer specification data store (REDIS) processes aerospike / REDIS / full index mongoDB / S3… incremental index web/file metasearch load balancer* document/ web links prometheus + monit alerts / monitoring performance logs document repositories * scraping & indexing infrastructure is fully scalable and distributed spot.my automatically scales to meet demand and is fully monitored with failure alerts

spot.my GPU enabled semantic search STAGE 1 (cluster articles) • Enables very fast searching of large Identify clusters of similar stories and corpora & vocabs with low RAM/CPU orphans (non-clustered) 27M papers > 5M clusters + 3M orphans query term GPU with fast GRAM + + x + + + + ++ + + STAGE 2 (cluster search) Find closest clusters and orphans GPU with fast GRAM STAGE 3 (results refinement) Full search of selected clusters to find hits CPU & cheap RAM

other spot.my features find relevant create subject iterate use keywords ‘like this’ - papers even if channels and searches to or whole drag & drop they use like/dislike get even paragraphs to whole papers different papers to better search as queries words refine matches

Creating Ne w Opportunities Much deeper insight into complex diseases • Novel (patentable) R&D / combinatorial interventions Includes genotypic, phenotypic and clinical data • Clinical trials design / patient stratification • Healthcare analytics / service planning Use of biomarker clusters in clinical decision support: • Personalized disease risk scoring and therapy selection • Personalized dietary and lifestyle advice