Recent advances in Rational Drug Repositioning ...

3 downloads 181 Views 57MB Size Report
May 9, 2017 - Director of Bioinformatics, Data Science and Precision Medicine .... Multiscale Biology, Icahn School of Medicine at Mount Sinai, Mount Sinai ...
Recent advances in Rational Drug Repositioning & Translational Bioinformatics Shameer Khader, PhD Director of Bioinformatics, Data Science and Precision Medicine Philips Healthcare, Cambridge, MA Adjunct member: Institute of Next Generation Healthcare & Dudley Laboratory Mount Sinai Health System, New York, NY [email protected] | @kshameer Novel Approaches to Understanding the Mechanisms of the Neuropsychiatric Symptoms in Alzheimer’s and Advancing Therapy Development Workshop Sponsored by the National Institute of Mental Health (NIMH) May 8-9, 2017

Acknowledgements

Design, development and analytics of RepurposeDB is supported by the following NIH grants: R01-DK09824203 and U54-CA189201-02

Public health relevance • Implementing data-driven methods that use real-time clinical variables in a hypothesis-free approach could help us to find new features • Designing predictive and prescriptive models would help to accelerate stratification of patients at risk for improved care • Major implications in quality of healthcare delivery and impact on patient outcomes • Precision medicine is poised to optimize care and reduce cost http://www.nejm.org/doi/pdf/10.1056/NEJMp1401111 http://jama.jamanetwork.com/article.aspx?articleid=1674245 http://www.thelancet.com/journals/langlo/article/PIIS2214-109X%2814%2970356-0/abstract Figures courtesy: The Lancet, JAMA and NEJM

Objective •

To discuss recent advances in translational bioinformatics



CTSA mission to design, develop and democratize biomedical and healthcare data science •

Rational Drug Repositioning



EMR-driven phenotyping and data harvesting



Phenotyping (Computational and algorithmic phenomics)



Data-driven disease subtyping



GWAS to PheWAS using EMR-linked biorepositories



Large-scale mining of disease and therapeutic trajectories Figure courtesy: B. Glicksberg

We live in a world of growing population, new diseases and resurgence of old diseases

Mental, behavioral and neurological combined = 10.4 DALY

Drug development is costly, market-driven and not scalable to address the population or patient-level challenge

Data is cheap, (almost readily) available and reusable

http://www.genome.jp/Fig/stats/db_growth.gif Nat Rev Mol Cell Biol 2009, 10(11):791-803.

Data-driven methods that combine computational and experimental approaches could complement, improve or reduce the cost of drug discovery Patient population that need better therapies

Drug repositioning

Druggable targets EHR-based validation

Molecular profiles

Biological networks

Experimental validation

Improved therapies to patients Families with rare disease

Disease modules

Figure from: Shameer et.al Current Topics in Medicinal Chemistry

Why drug repositioning? Bring therapies to market in approximately half the budget & time required by traditional drug development cycle.

Figure from: Shameer et.al Current Topics in Medicinal Chemistry

Systematic or Target based drug repositioning? How to select a drug repositioning strategy?

Figure from: Shameer et.al Current Topics in Medicinal Chemistry

Icahn School of Medicine at Mount Sinai, New York, USA. +Currently at Philips Healthcare, Cambridge, USA

Introduction

Gene Signature-based Repurposing

Multiple factors such as pharmacogenetics, comorbidities, and side effect tolerance influence patients’ clinical responses to statin therapy. Furthermore, it has been demonstrated that statins are biologically pleiotropic agents which effect a variety of cellular functions in addition to their characteristic effect of HMG-CoA reductase inhibition. Previous work has demonstrated an unexpected pravastatin-paroxetine adverse interaction of increased blood glucose. 1

Rational drug repositioning and validation using RWE from EMR

Gene Signature-based Repurposing The recent Yellow-II trial by Kini et al.2 used a combination of optical coherence tomography (OCT), near infrared spectroscopy (NIRS), and intravascular ultrasound (IVUS) intravascular imaging with network biology to characterize changes in plaque morphology and transcriptomics in response to high-dose rosuvastatin therapy. In this study, 117 differentially expressed genes were detected from baseline to follow up, with a variety of roles including cholesterol metabolism, collagen catabolism, platelet function, and especially the immune system. Here, we explore their potential utility for drug target discovery and drug repurposing.

Drug Repurposing

R

Postive Correlation

Yellow II Gene Expression Gene 1

Original Scale

Negative Correlation

Gene n

Drug Activity

• Drug repurposing is the application of known, previously approved drugs for new disease indications

Negative Correlation with Transcriptomic Signature Positive Correlation with Transcriptomic Signature

• Drug repurposing discovery is generally categorized as: a) Disease-signature based

Y = Xβ + Ζγ + ε • Fixed effects for demographic covariates and drug administration • Random intercept per subject • Lab values evaluated before and after drug administration

Postive Correlation

b) Target based Acta Crystallogr D Biol Crystallogr 63(Pt 1): 80-93 Jan 2007

Results

J Mol Bio 426(4):843852 Feb. 2014

Yellow II Gene Expression Gene 1

• Evaluation of our #2 predicted drug in conjunction with any statin Drug

Imatinib (small molecule)

Negative Correlation

PCSK9 inhibitor (mAb)

Aims Drug Activity

Gene n

Negative Correlation with Transcriptomic Signature

Here, we aimed to use Yellow-II transcriptomic results to: 1. Explore five detected novel genes which have limited biochemical evidence, functional annotations, or previous associations with disease 2. Use transcriptomic results for the discovery of potential drug targets and candidates drugs which may be repurposed for adjuvant statin therapy

Positive Correlation with Transcriptomic Signature

Transcriptome Analysis

Keyword Cluster Cholesterol Metabolism SREBP Lipid Synthesis Control Innate Immune Response SH3 Domain Platelet Activation

Gene Count 7/105 3/17 7/105 6/105 5/97

47,258

31,026

30,452

Discovered Drug

6,354

2,818

2,710

New Drug + Statin

1,801

1,182

1160

Demographics Age (years) Sex (%) Ethnicity (%)

Fold Enrichment 24.1 35.8 2.4 5.5 4.8

Mean Std. Dev. Male Female Caucasian African American Asian Hispanic/Latino Other

All 64.35 12.6 58.25 41.75 16.01 10.61 1.55 13.21 58.63

Statin 64.67 12.35 58.74 41.26 16.11 10.19 1.59 12.88 59.24

New Drug 62.54 14.25 50.48 49.52 15.02 16.75 0.74 20.26 47.23





Both 62.57 13.67 48.80 51.20 13.20 9.6 0.0 19.20 58.0







Laboratory Test

• 5 of 117 genes had limited previous functional annotation Protein Observed Experimental Network Relationships3 TMEM135 Co-localization; (18) Genetic interactions; (5) Physical interactions; (1) TMEM51 Co-localization; (19) Physical interactions; (1) SCIMP Co-expression; (24) Physical interactions; (2) KIAA1551 Co-localization; (18) Genetic interactions; (1) Physical interactions; (2) C15orf39 Co-localization; (19) Genetic interactions; (2) Physical interactions; (1)

Prescribed Drug With Lipid Labs With Labs & Demographics

Any statin

Y = Xβ + Ζγ + ε

Physically Interacting Genes3 CDC23 MRPL12 LYN, RGS16 TBC1D4, EXOC1 RPLP1

• Fixed effects for demographic covariates and drug administration • Random intercept per subject • Lab values evaluated before and after drug administration

Total Cholesterol LDL HDL

1. 2. 3.

Triglycerides LDL:HDL Cholesterol:HDL

Laboratory Testet.al Statin Unpublished data Kipp Total Cholesterol

n School of Medicine at Mount Sinai, New York, USA USA. +Currently at Philips Healthcare, Cambridge, USA

Effect of the drug and in conjunction with statin on various CVD biomarkers

ed Repurposing

Results Original Scale

Log Transformed

orrelation with omic Signature

Y = Xβ + Ζγ + ε • Fixed effects for demographic covariates and drug administration • Random intercept per subject • Lab values evaluated before and after drug administration

s

ion with any statin

abs With Labs & Demographics 30,452

Laboratory Test

Drug Effect

β Coefficient P Val

Total Cholesterol LDL

-3.82 mg/dL -3.81 mg/dL

0.0030 0.0004

HDL

+1.87 mg/dL

80% 95,373 drug pairs were enriched for coadministration Drug A

&



AA individuals had more prescription instances and unique medications compared to all other racial groups



HL individuals had the fewest prescription instances and unique medications compared to all other groups



Polypharmacy (4+ simultaneous drug prescriptions) varied according to race (χ2 p

Suggest Documents