PrecisionMedicineinRareDiseases:

AcrossCon2nentsandDisciplines

Matthias Kretzler MDDiv. Nephrology / Internal Medicine

Computational Medicine and Bioinformatics

University of Michigan Medical School

The challenge for target identification in rare disease

•  Descriptive disease categorization into syndromes with multiple pathogenetic mechanisms–  Problems of ‘mixed bag’ diseases:

•  Unpredictable disease course and response to therapy•  Medicine as an ‘art of trial and error’

‚DieUrinbeschau‘UnknownmasterLowerRhinevalley,ca.1760

Description of

Renal Failure

anno 1760

•  Descriptive disease categorization into syndromes with multiple pathogenetic mechanisms–  Problems of ‘mixed bag’ diseases:

•  Unpredictable disease course and response to therapy•  Medicine as an ‘art of trial and error’

•  Shift in our disease paradigms:–  Mechanism based personalized patient management

•  Define the disease process active in the individual patient

–  Define molecular disease causality–  Base prognosis on specific disease process–  Target therapy to interfere with the mechanism

currently destroying organ function

The challenge for target identification in rare disease

The GOAL:Precision Medicine for Rare Disease

(modifiedfromMarianiandKretzler,NDT,2015)

The advantage of being a Nephrologist:Biopsy centred clinical and molecular phenotyping

Cohenetal.,2002

Research networks for molecular analysis of renal disease

Standardizedprotocolimplemented:>3800biopsiesprocuredandprocess,>13000paSentsincohorts

ERCB

Cohorts for molecular analysis of CKD2560 glomerular disease biopsies for gene expression studies24 European Centers

1165 Nephrotic Syndrome Patients in Registry708 incipient Nephrotic syndrome Cohort23 Centers in North-America

1200 CKD patients in 7 US-Centers in Mid-West

H3 CKD Africa, 8400 patients and controls in Sub-Saharan Africa with biosamples and prospective follow-up

180 Diabetic Nephropathy Cohort in Pima Native American Interventional Trial 98 protocol biopsies available for molecular analysis

760  IgA-Nephropathy Cohort at 1 PKU, Beijing, ChinaProspective biosamples and follow-up

680 incipient Nephrotic Syndrome Cohort

Molecular disease

definition:

Why do I have this disease?

Systems genetics concepts

Genetic variants affect

regulatory and proteomic machinery of the cell,

leading to disruption in a metabolic pathway

resulting in clinical trait / renal

disease phenotype.

B. Keller et al. Kidney Int, 2012

Integration SNPs – transcripts- clinical traits

B. Keller et al. Kidney Int, 2012

GWASCKDGen-candidatemarkers

1.FiltercandidategenesforGFRcorrelatedmRNAexpression

2.DetecSonofco-regulatedtranscriptsforeverycandidategene

3.DetecSonofenrichedpathwaysamongco-regulatedtranscripts

4.DetecSonCKD-associatedpathwayinterdependencies

SystemsGene*csofGlomerularDisease

Pathway network associated with shared CKDgen co-regulated transcripts. Spring embedded algorithm. Node size reflects degree of connectivity, edge thickness increasing with more genes shared among two pathways. Node color reflects number of transcript associated with pathway: (multiple=red, few=green)

Martini et al. J Am Soc Nephrol, 2014

SystemsGene*cs:SharedPathwaysinGlomerularFailure

Metabolism

Xenobiotic Metabolism SignalingAryl Hydrocarbon Receptor SignalingPPARgamma signalingPXR/RXR ActivationLPS/IL-1Mediated Inhibition of RXRFunctionTryptophan Metabolism…

Inflammation – Stress response

NRF2-mediated Oxidative Stress ResponseCdc42 SignalingNF-kappaB SignalingDendritic Cell MaturationJak/Stat SignalingCD28 Signaling in T helper Cells…

Definition of highly interconnected nodes (clusters)

FSGS pathways

Lupus Nephritis pathways

Differentially regulated genes were enriched in pathways indicated by the blue color.

Diabetic Nephropathy pathways

Early DN

Tub

uloi

nter

stiti

um c

ompa

rtm

ent

Progressive DN

Glo

mer

ular

com

part

men

t

Differential regulation Down-regulatation

Up-regulation

Control (LD+MCD): n=12 Early DN: n=24

Control (LD+MCD): n=8 Prog. DN: n=7

Control (LD+MCD): n=7 Prog. DN: n=11

Control (LD+MCD): n=11 Early DN: n=22

Cytoplasm

Nucleus

Cytoplasm

Nucleus

Cytoplasm

Nucleus

Cytoplasm

Nucleus

SharedChronicInflamma*oninCKDJak-Statpathwayac*va*on

Berthier et al., Diabetes. 2009;58:469

Repurposing into Glomerular Disease: Phase II trial of Jak 2 inhibition

BariciSnib,oralJAK2inhibitor:efficacyinRAandPsoriasis =>RepurposedinDiabeScNephropathy

•  RCTbyEliLillyinDN•  Primaryoutcome:ChangeinurineACRfrombaselineto24week

0 .6

0 .8

1 .0

1 .2

1 .4

T r e a tm e n t

Fo

ld c

ha

ng

e f

rom

ba

se

lin

e

P lo t o f L e a s t S q u a re M e a n s + /- S ta n d a rd E r ro r fo r 2 4 -h o u r U A C R

3 m o n th s 6 m o n th s 4 w k s w a sh o u t

M ix e d m o d e l re p e a te d m e a s u re s a n a ly s is o f lo g - tra n s fo rm e d d a ta w ith re s u lts b a c k tra n s fo rm e d .*p -v a lu e < 0 .0 5 ; * *p -v a lu e < 0 .0 1 b a s e d o n tre a tm e n t d iffe re n c e c o m p a re d to p la c e b o .

**

** **

P la c e b o 0 .7 5 m g Q D 0 .7 5 m g B ID 1 .5 m g Q D 4 m g Q D

* *

26 19 23 22 22 24 17 23 23 19 26 20 24 22 21

n -va lu e s :

Albuminuria Response Urinary IP-10: Target Engagement

From target identification to phase II completion in 42 months

Molecular disease

definition:

What will this

disease do to me?

Predic*onofCKDProgressionSelec*onofpa*entsatrisk

RiskPredicSon: –  Use outcome to select predictor from genomic data set

A set of tissue based candidate

markers

Validation markers in independent cohorts

Validated markers for testing in body fluids

Further validation in larger cohort

From*ssuetourine:Non-invasiveBiomarkersforCKDprogression

Enrichment analysis: signaling pathways & Therapeutic targets

Treatmenttargets

Biomarkers

Ju et al, Science Trans Med, 2015.

Table X: Top 10 upstream regulators of eGFR slope correlated genes. Upstream Regulator

Molecule Type

p-value of overlap

Target molecules in dataset

EGF growth factor

2.09E-12 ACTN1,APOA1,B4GALT5,CCL20,CCND2,CD44,CDC42EP1,CLDN3,CLU,CTSD,DUSP6,ELF3,HIF1A,ICAM1,ID2,IER3,ITGB3, LCN2,LTF,MYC,MYCN,NCAN,NOS2,NRP1,PBK,PPIB,RRM2,SERPINA3,SLC37A1,SNAP25,SOX4,SPP1,SYP,TFPI2,THBS1, VCAN,VIM

TP53 transcription regulator

4.83E-11 ACTN1,ALOX5,ANLN,ANTXR1,ANXA1,APAF1,APOA1,ASF1B,ATL3,CCL2,CCND2,CD44,CLU,COL4A1,CTSD,CXCL1,DLGAP1, EIF4G3,ELF4,FHL1,FHL2,GDF15,GLIPR1,HIF1A,HS3ST1,ICAM1,ID2,IER3,ILK,KITLG,KRT18,MCM7,MET,MYC,MYOF,NOS2, NRP1,PBK,PDLIM1,PFN1,PMEPA1,PPFIBP1,PPIC,PTPRE,RRM2,SGPL1,SIVA1,SOD2,SPHK1,SPP1,TFPI2,TGFB2,THBS1, THBS2,TMSB10/TMSB4X,TNFSF9,TOP2A,TPX2,TSC22D3,UBE2C,VCAN,VIM,XRCC5,ZMAT3, ZYX

IL1B cytokine 3.45E-10 ADAMTS1,ANTXR1,ANXA1,CCL2,CCL20,CD44,CD74,CFB,CXCL1,CXCL6,ELF3,FAM129A,G0S2,GDF15,HIF1A,HMGA1, HSD11B1,ICAM1,IER3,ITGB3,ITGB8,LAMC2,LCN2,LHB,MARCKSL1,MMP7,MYC,NFKBIZ,NOS2,NRP1,OSM,PIGR,PSMB10, SERPINA3,SNAP25,SOD2,SOX9,SPP1,TAC1,TFPI2,THBS1,TSC22D3,UGCG,VCAN,VIM,ZYX

TGFB1 growth factor

1.36E-09 ACTN1,ALOX5,CCL2,CCL20,CCND2,CD207,CD44,CFL1,CLU,COL4A1,COL4A2,CTSD,CXCL1,DARC,DYRK2,ELF3,ELF4, FCER1A,FHL1,FNDC3B,FOSL2,GDF15,GGT6,GNG7,HIF1A,HMGA1,ICAM1,ID2,IER3,ILK,ITGB3,ITGB6,KCNG1,KDELR3, KDM5B,KITLG,KLF9,KRT18,LAMC1,LAMC2,LCN2,LOXL1,LOXL2,MET,MMP7,MYC,MYCN,MYOF,NCAN,NNMT,NOS2,NRP1, OSM,PDLIM7,PKIG,PMEPA1,PNMT,SERPINA3,SOD2,SOX4,SOX9,SPARCL1,SPHK1,SPP1,SYP,TGFB2,THBS1,TOP2A, TSC22D3,TUBA1A,VCAN,VIM,ZYX

IL6 cytokine 1.28E-08 ADAMTS1,ANXA1,APOA1,ARL4C,CCL2,CCL20,CD74,CLU,CXCL1,CXCL6,DUSP6,HIF1A,ICAM1,ID2,ITGB3,JAK1,KRT18, LCN2,LTF,MET,MPO,MYC,NOS2,PSMB10,REG1A,RRM2,SENP2,SERPINA3,SOD2,SPP1,STS,SYP,TAC1,THBS1,TOP2A, UBE2C,VIM,XRCC5

TNF cytokine 2.54E-08 ALOX5,ANXA1,APAF1,APOA1,CCL2,CCL20,CCND2,CD44,CFB,CLCF1,CLU,CXCL1,CXCL6,DUSP6,ELF3,FOSL2,G0S2, GDF15,HIF1A,HSD11B1,ICAM1,IDE,IER3,IFNAR2,ITGB3,ITGB6,JAK1,KITLG,LAMC1,LAMC2,LCN2,LGALS12,LIPE,MARCKSL1, MET,MMD,MMP7,MPO,MYC,NCAN,NFKBIZ,NNMT,NOS2,NRP1,OSM,PIGR,PSMB10,RRM2,SERPINA3,SOAT1,SOD2,SOX4, SOX9,SPHK1,SPP1,TAC1,TAPBP,TFPI2,TGFB2,THBS1,THBS2,TNFSF9,TSC22D3,UGCG,VIM,ZYX

CSF2 cytokine 9.96E-08 ALOX5,ANLN,ANXA1,APAF1,CCL2,CD63,CD74,CLCF1,CXCL1,DUSP6,GDF15,ICAM1,ID2,IER3,ITGB3,MET,MMD,MPO,MYC, NOS2,NRP1,OSM,RRM2,SOD2,SPP1,THBS1,TOP2A,TPX2,UBE2C

CEBPA transcription regulator

1.13E-07 ANXA1,ARL4C,CCL20,CCND2,FHL1,G0S2,GLIPR1,HSD11B1,ICAM1,ID2,IER3,LCN2,LTF,MALT1,MPO,MYC,MYF5,NRP1, PTPRE,SOD2,SPP1,TAC1,TGFB2,TRIB1,TSC22D3,VCAN

NFkB complex 1.48E-07 CCL2,CCL20,CCND2,CD44,CD74,CFB,CLCF1,CLU,CXCL1,CXCL6,ELF3,G0S2,GDF15,HGFAC,HIF1A,ICAM1,IER3,IFNAR2, ITGB8,LCN2,MYC,NFKBIZ,NOS2,SENP2,SERPINA3,SOAT1,SOD2,SPP1,TAC1,TFPI2,TRIM38,VIM,XRCC5

ERBB2 kinase 3.46E-07 CCL20,CCND2,CLDN3,COL4A1,DHH,DUSP18,DUSP6,ELF3,ETV6,HIF1A,ID2,ILK,ITGB3,KCNN4,KDM5B,LAMC2,LCN2,MYC, MYCN,NNMT,NOS2,NRP1,PMEPA1,RRM2,SERPINA3,SOX4,SPARCL1,SPINT1,TAPBP,THBS1,TOP2A,TUBA1A,UBE2C, VCAN,VIM

EGF,topup-streamregulatorofrenallosscorrelatedgenes

uEGF: Hazard Ratio for time to event in glomerular disease

PKU-IgAN 0.53 (0.37-0.69)

NEPTUNE 0.29 (0.15-0.58)

C-PROBE 0.27 (0.13-0.54)

Cohort HR (95% CI)

Adjusted HR of uEGF/Cr

0.0 0.5 1.0 1.5

PKU-IgAN 0.53 (0.37-0.69)

NEPTUNE 0.29 (0.15-0.58)

C-PROBE 0.27 (0.13-0.54)

Cohort HR (95% CI)

Adjusted HR of uEGF/Cr

0.0 0.5 1.0 1.5

PKU-IgAN

NEPTUNE

C-PROBE

Cohort

Adjusted HR of uEGF/Cr

0.0 0.5 1.0 1.5

A

B

C

Mul*variable-adjustedhazardra*osurinaryEGF/Crforoutcomes.HRsadjustedbyage,gender,eGFRandACRandobtainedbyindependentCoxregressionmodelsineachstudycohort.

Molecular disease

definition:

What can be done to help

me?

DefiningMolecularSubgroupsinGlomerularFailure

Define functional disease group => associate with outcome => predictors of group

CR*

Cluster1 Cluster2 Cluster3

GFR* 108mL/min/m2 79 42

UPCR 0.9 1.25 1.7

MarianiL,MarSniS,NairV,….

FSGS-MCDRenalBiopsygeneexpressioncluster

Cluster1 Cluster2 Cluster3

GFR 104 94 69

71/202genessign.Upregulated–concordantwithNEPTUNEgenesthatdifferenSatedcluster3there(125/131alsoupregulated)

ValidaSoninindependentERCBcohort

MarSniS,CohenC,NairV

ERCB

The Path Forward:Building a Precision Medicine Research Community

(modifiedfromMarianiandKretzler,NDT,2015)

Sharing Knowledge in the Informational Commons

•  Disease community specific low threshold outreach tools

•  Nephroseq: Kidney specific web based search engine

Combined data base and systems biology search engine for standardized analysis by the renal research community

DataCommonsforRareDiseaseResearchNetworks:TranSMARTDataIntegra*on&Analysis

The Challenge:•  Data Integration & Analysis

Athey and Omenn, 2009

A solution:•  tranSMARTOpen-sourcesoluSonforsharing,integraSon,standardizaSonandanalysisofheterogeneousdatafromcollaboraSvetranslaSonalstudiessupportedbyanopen-databiomedicalresearchcommunity

Overviewathip://lanyrd.com/2014/transmart/sdfqkf/

Making Vision Reality :Consortiums and Collaborative Science

in Rare Renal Disease

RPC2

RenalPrecompeSSveConsorSum

MolecularTargetDefiniSoninCKD

Sharedontology+biosamplingprotocolinindependentlygovernedcohortsusingindividualtranSMARTinstances:

De-idenSfieddataelementscaneasilybeshared,compared,andinterpretedamongnetworks

NEPTUNE

NorthAmerican

NephroScSyndrome

(FSGS/MCD/MN/IgAN)

CureGN

North-AmericanGlomerularDiseases

(IgAN,MN,FSGS,MCD)

ERCB

EuropeanCKD(DKD,HTN,LN,IgAN,FSGS,MN,MCD,VasculiSs)

NEPTUNEChina

ChineseNephroSc

Syndrome(MN,FSGS,MCD)

SystemsBiology ClinicalData

ModelSystems

NewTargetsandBiomarkers

BioinformaScsAdvisoryCommiiee

MindtheGap:RenalPre-Compe**veConsor*um(RPC2)for

moleculartargetiden*fica*on

Precision Medicine in Rare Disease

translate such insights into novel therapies (6). The success-ful use of whole-exome sequencing to identify the geneticbasis for response to cytotoxic T-lymphocyte antigen 4blockade in melanoma is one such example (7). In this land-mark study, a discovery cohort of only 25 patients (11 whoresponded to therapy and 14 who did not) was analyzedthrough a bioinformatic pipeline, which included sequenc-ing for mutational load in the tumor tissue, translatingmutations into altered peptides, and simulating MHC class1 binding, leading to the identification of a neoepitope sig-nature that correlated with clinical response. Furthermore,the Cancer Genome Atlas (TCGA) aims to establish a coreresource to facilitate this approach in other cancer diagno-ses. TCGA is funded by the National Cancer Institute andthe National Human Genome Research Institute and hascollected tumor and normal tissue from 11,000 patientswith 33 types of cancer to generate multiple large–scaledatabases of DNA, RNA, and proteomics. Among otheradvances, this database was used to identify biologicallybased subtypes of glioblastoma, setting up improved trialdesign, and enabling novel therapeutic developments (8,9).This article will highlight definitions, important limita-

tions, and approaches to large–scale dataset generation inglomerular disease (excluding genetics, which will be dis-cussed separately in this series). Examples of the successful

application of an integrative approach in glomerular dis-ease will be provided, with the anticipation that more suc-cess stories are expected in the future. The power ofanalyzing the human disease as a unified whole has inde-pendent value, but combining with parallel work in modelsystems, cell cultures, and observational and interven-tional clinical studies allows each approach to inform theother and expedite discovery and clinical translation.

Generation of Large-Scale Data in GlomerularDiseaseGeneration of multilayered large–scale datasets for glo-

merular disease begins to address the goals of precisionmedicine for nephrology (Figure 1). Not unlike oncology,there are unique advantages of applying this approach tokidney disease, where the biospecimens needed to gener-ate the datasets highlighted in Figure 1 are frequently ob-tained during routine clinical care. For example, biopsytissue (morphology) not only provides valuable structuralinformation currently in clinical use, but also offers a win-dow into the molecular mechanisms active in the tissue ofthe affected organ (transcriptome). Another advantage isthe easy access to urine, which can be tested repeatedlyand serves as a potential liquid biopsy to capture changes

Figure 1. | Large–scale dataset integration across the genotype-phenotype continuum. Integrative biology approaches can be initiated withany dataset across the genotype-phenotype continuum and then, evaluated against the other domains to help identify novel risk factors,pathogenic pathways, or biomarkers of progression and therapeutic response. As a result, novel mechanisms can be further investigated inmodel systems to confirmfindings and test therapies.Mechanistic–based patient groups can be identified and targeted for enrollment in proof ofconcept clinical trials.

2 Clinical Journal of the American Society of Nephrology

Mariani,Pendergran,Kretzler,cJASN,2016

•  Independently governed, jointly designed cohorts for mechanistic patient stratification

•  Mechanistic disease pathways cutting across multiple rare disease mapped (NCATS-RDCRN)

=> Risk based disease stratification=> Molecular target identification

•  Repurposing successfully implemented and scalable into adequately targeted patient populations

Team science of translational medicine

ERCBGrantsupportby