medically refractory ulcerative colitis: gwas to translation · many ulcerative colitis patients...

$: Medically Refractory Ulcerative Colitis: GWAS to Translation · Many Ulcerative Colitis patients are refractory to medical therapy Inflammation cannot be controlled by “mild”$
Medically Refractory Ulcerative Colitis:

GWAS to Translation

Talin Haritunians Medical Genetics Institute

Cedars-Sinai Medical Center

Ulcerative Colitis

Current state of genetics of UC

In a perfect world, how can GWAS

translate to the clinic?

GWAS on Medically Refractory

(medically “unresponsive”) UC

Identifying “at risk” patients

Crohn’s Disease (CD)

Transmural Inflammation

Ileitis Ileocolitis Colitis

Ulcerative Colitis (UC)

Mucosal ulceration in the colon

Symptoms include vomiting, diarrhea, weight loss and abdominal pain.

Hippocrates (~400BC) first description of

fistulae

“Ulcerative Colitis” defined as distinct from

infectious dysentery and nervous diarrhea

Wilks (1859)

Electric sigmoidscopy for proper diagnosis

Careful study of clinical features brought the

recognition that Ulcerative Colitis was a unique

disorder

Fenwick (1889); Dalziel (1913)

Case classification in British Medical Journal

Hawkins presented UC characteristics drawn from

85 cases collected throughout Great Britain (1909)

Careful clinical observation to note great differences in

length of “flares”

Amount of

inflammation

of the lining of

the colon

Amount of

inflammation

of the lining of

the colon

Length of colon

affected

Higher risk for family

members

Higher risk for monozygotic

twins

MZ > DZ = 1st degree

relatives >> spouses

Higher risk for some ethnic

groups (AJ)

lower risk for others (Japan)

higher in whites, lower in

blacks in South Africa

“Knockout” of several

different genes in mice causes

intestinal inflammation

First DegreeRelative

Jews BlacksJ

> 10x

~ 4x

~ 4x

Ris

k o

f IB

D

First

Degree

Relative

with IBD

NJ

Caucasian

AA

Genetic susceptibility of patient

Mouse model knock-outs demonstrate that many genes

are possible candidates for UC

GWAS confirmed 47 UC associated loci to-date

Gut microbiota

Intestinal inflammation does not occur when mouse

models of UC are raised in a germ-free environment

IBD patients have antibodies to microbial components

Dysregulated immune response to commensal gut

bacterial flora

Transfer of various combinations of inflammatory and

regulatory T-cells increase or decrease intestinal

inflammation in mouse models

Aminosalicylates

Mild to moderate inflammation

Corticosteroids

Moderate to severe inflammation

Non-responders to aminosalicylates

Immunomodulators

Severe inflammation

Can take 6 months for full benefit

Surgery

~20% of patients

Meta-analysis of 6 GWAS

Cohort Population UC cases Controls

Cedars-Sinai LA Caucasian 723 2,880

German German 1,036 1,694

CHOPSTICKS CHOP Caucasian 643 6,197

NIDDK North America 977 2,122

Swedish Sweden 948 1,408

WTCCC UK 2,360 5,417

Total 6,687 19,718

Many Cohorts

(10+)

Population UC cases Controls

Total European 9,628 12,917

Discovery

Replication

Anderson et al 2011. Nature Genetics 43:246-252.

GWAS analysis:

6,687 cases & 19,718 controls (1.1M SNPs)

18 previously confirmed susceptibility loci

Follow-up analysis:

9,628 cases & 12,917 controls

50 novel susceptibility loci (p<1x10-5)

29 of 50 novel loci

confirmed(p<5x10-8 combined

analysis)

18 loci

~11% heritability

47 loci

~16% heritability


Observ

ed -

1[log(p

)]

Expected -1[log(p)]


Of 47 loci:

3 regions contain single gene

35 contain multiple genes

9 gene deserts

Multiple genes associated

with:

Cytokines, cytokine

receptors & key regulators of cytokine-mediated signaling

Innate & adaptive immune response

Macrophage activation

Apoptosis regulation

Intestinal barrier function (tight junction assembly in epithelial

cells)

GNA12 Guanine nucleotide

binding protein alpha12

Membrane bound GTPase with important role

in tight junction assembly in epithelial cells

FCGR2A/B FcG receptor Transport of antigen complexes in response to

gut bacteria

CARD9 Caspase recruitment

domain 9

Adaptor molecule required for response to

pathogens, including fungi

TNFRSF14 TNF receptor

superfamily

T-cell transfer model of colitis, expression in

innate immune cells plays important role in

preventing intestinal inflammation

TNFRSF9 TNF receptor

superfamily

Involved in regulation of peripheral T-cell

activation and is expressed at sites of

inflammation

IL23R IL23 receptor Differentiation of TH17 cells

IL7R Receptor for IL7, key

regulator of naïve &

memory T-cell survival

Increased IL7R expression in T-cells in human

& murine colitis; Selective depletion of these

cells ameliorates established colitis

Many Ulcerative Colitis patients are refractory to

medical therapy

Inflammation cannot be controlled by “mild”

therapies (aminosalicylates) nor “stronger”

therapies (IV corticosteroids, cyclosporin, or anti-

TNF biologics)

Why should we identify patients at risk of

medically refractory Ulcerative Colitis (MR-UC)?

Early introduction to more intensive therapy

Shorter time to surgery

Reduction in patient suffering

Aminosalicylates


Corticosteroids



Immunomodulators

Severe inflammation


Surgery

20% of patients

Aminosalicylates


Corticosteroids



Immunomodulators

Severe inflammation


Surgery

20% of patients

✗

✗

✗

GWAS comparing medically responsive

& medically refractory UC (MRUC)

patients

UC requiring colectomy for symptoms

uncontrolled by medical therapy

Identify SNPs that predict UC patients

that did not respond to medical

therapies


& medically refractory UC

(MRUC)patients





therapies

FACTORS

Responsive

UC

(n=537)

Refractory

MRUC

(n=324) P

Sex (F%) 47% 47% 0.89

Median Age of UC Onset (yrs) 26 27 0.93

Extraintestinal Manifestations (%) 19% 15% 0.16

Smoking (%) 8% 6% 0.24

Median Disease Duration (mo) 95 48 7.4x10-9

Extensive Disease (%) 64% 80% 2.7x10-6

Family History of UC (%) 15% 24% 0.004

Haritunians et al 2010. IBD 16:1830-40.

Genotyping Data

All samples genotyped with Illumina CNV370K chip

Samples with high rate of genotyping retained (>98%)

313,720 SNPs passed quality control

MAF >3%; HWE ≤ 0.001; SNP failure rate <10%

No differences in SNP missing data between cases and

controls

Principal component analysis used to adjust for

population stratification (Eigenstrat)

Association tested with Logistic Regression corrected

for 20 principal components (PLINK, R)

Statistical method for handling datasets with

large number of measurements (dimensions)

by reducing these dimensions to a few PC that

explain the main patterns observed

Corrects for population stratification (the

difference in allele frequencies between cases

and controls due to ancestral differences) that

may lead to spurious association in GWAS

Component 2

Com

pon

en

t 1

MR-UC Non-MR-UC

“Caucasian Axis”

-log10(P

-valu

e)

2 1 3 4 5 6 7 8 9 10 11 12 13 14 15 16 22

21 20

19 18

17

Chromosome

TNFSF15

MHC

-log10(P

-valu

e)

2 1 3 4 5 6 7 8 9 10 11 12 13 14 15 16 22

21 20

19 18

17

Chromosome


82 SNPs p<1x10-3, with

10 SNPs reaching

genome-wide significance

Association peak at 10-16

MHC association with

severe UC confirms

earlier established

association of MHC with

UC


17 SNPs p<1x10-3, with 3 SNPs

reaching genome-wide

suggestive significance

Association peak at 10-6

Increased expression observed

in inflamed mucosa of colon

and small bowel (CD)

Increased expression also

correlated with severity of ileal

and colonic inflammation in

mouse model

Neutralizing antibodies prevent

and treat established chronic

intestinal inflammation


& medically refractory UC

(MRUC)patients





therapies

Each SNP was independently analyzed

Contribution of top SNPs to MR-UC was tested using forward logistic regression and Cox proportional hazards on time-to-surgery data (R)

MR-UC vs. Non-MR-UC

(n=324) (n=537)

Top 100 SNPs

(p < 3x10-4)



MR-UC vs. Non-MR-UC

(n=324) (n=537)

Top 100 SNPs

(p < 3x10-4)

37 SNPs



MR-UC vs. Non-MR-UC

(n=324) (n=537)

Top 100 SNPs

(p < 3x10-4)

37 SNPs

MR-UC (<60 mo; n=187)

vs. Non-MR-UC (n=328)

Top 65 SNPs

(p < 1x10-4)

9 SNPs



MR-UC vs. Non-MR-UC

(n=324) (n=537)

Top 100 SNPs

(p < 3x10-4)

37 SNPs

MR-UC (<60 mo; n=187)

vs. Non-MR-UC (n=328)

Top 65 SNPs

(p < 1x10-4)

9 SNPs

46 SNPs explaining 47.6% of risk for MR-UC (p-value Cox < 10-16)

+



9 rs7861972 6759692 JMJD2C | SNRPEL1 |

9 rs3118292 25133480 9p21.3

9 rs10817934 118589872 ASTN2

11 rs2403456 11134390 11p15.3

11 rs1461898 37546808 11p12

11 rs6591765 62674829 SLC22A24 | SLC22A25 | SLC22A10 |

12 rs887357 3344906 12p13.32

12 rs526058 24326688 12p12.1

13 rs7319358 78448935 13q31.1

14 rs1956388 28202628 14q12

14 rs11156667 30906111 GPR33 | HEATR5A | NUBPL | C14orf126 |

14 rs8020281 94436179 14q32.13

14 rs10133064 85844148 14q31.3

Risk Score as Total Count of Risk Alleles

Score is the total number of risk alleles (each of the 46 SNPs may contribute 0, 1, or 2)

Range of Observed Risk Score 28-60 Possible Range 0-92

Higher scores associated with MR-UC

108 309 69 0

1 64 199 50

28-38 39-45 46-52 53-60

MR-UC

NON-

MR-UC

SCORE

0

20

40

60

80

100

28-38 39-45 46-52 53-60

Higher Risk Score Associated with MR-UC

28-38 39-45 46-52

Risk Score Categories

Pro

port

ion

(%

)

Non-MR-UC

MR-UC p-value Chi-squared test for trend < 10-16

53-60


Sensitivity & Specificity of Risk Score Model Following

10-fold Cross-Validation Indicate the Model was Robust

Sensitivity/ Specificity (cut-off=0.5)

Original Data with logistic regression

-----------------------------------------------------------------

0.858 Specificity

0.793 Sensitivity

-----------------------------------------------------------------

1000 times of 10 fold Cross-Validation data sets with logistic regression

--------------------------------------------------------------------------------------------------------------------

0.859 0.0021 0.858 0.870 1000 Specificity

0.789 0.0067 0.758 0.793 1000 Sensitivity

--------------------------------------------------------------------------------------------------------------------

Mean Std Dev Minimum Maximum N Variable

--------------------------------------------------------------------------------------------------------------------


Detecting “true” positives

How specific is my test for particular disease?

Time to Surgery (months)

Cu

mu

lati

ve P

robabilit

y o

f

Avoid

ing C

ole

cto

my (M

R-U

C)

39-45; n=373

53-60; n=50

46-52; n=268

28-38;

n=109


Cu

mu

lati

ve P

robabilit

y o

f

Avoid

ing C

ole

cto

my (M

R-U

C)

60

39-45; n=373

53-60; n=50

46-52; n=268

28-38;

n=109



Cu

mu

lati

ve P

robabilit

y o

f

Avoid

ing C

ole

cto

my (M

R-U

C)

60

39-45; n=373

53-60; n=50

46-52; n=268

28-38;

n=109

24

8.3%

48.4%

84%

0%

19.1%

62%


Chr SNP Loci Pathway

6 rs17207986 TNXB Extracellular adhesion

6 rs3734263 UHRF1BP1 T-cell death

6 rs9470224 MAPK13,14 TCR & TLR signaling pathway

9 rs11554257 TNFSF15 Th1-Th2-Th17 activation

12 rs1144720 BICD1 Chlamydia inclusions; Antigen processing

17 rs11891 CANT1 Signal transduction

19 rs11085825 DNASE2 Lysosomal function

CALR Antigen processing

19 rs4808408 CYP4F2 Leukotriene pathway

20 rs6039206 PLCB1 Signal transduction

20 rs6059101 C20orf186 Antimicrobial peptide cluster

BICD1

DNASE2

C20orf186

Epithelial cell

Innate/ Myeloid

T cell activation

T cell

Cytokine signaling

UHRF1BP1 CALR TNFSF15

MAPK13,14 CANT1 PLCB1

TNFSF15

TNFSF15

Potential

Therapeutic

Targets: Genes

from MR-UC

analyses

GWAS on 861 UC patients confirmed major contribution of MHC region to UC severity

SNPs identified by GWAS may together explain a large proportion of risk 46 SNPs ~48% risk of MR-UC requiring

colectomy

Each SNP made a small contribution (OR 1.2-1.9)

Combination of risk alleles may be useful to predict medically refractive UC

Identified interesting pathways for further investigation of potential new therapeutic targets in MR-UC

YES

“Skip” to more aggressive therapies with

higher SNP scores.

Cedars-Sinai Medical Center

Medical Genetics

Institute: Jerome Rotter

Kent Taylor

Talin Haritunians

Xiuqing Guo

Emebet Mengesha

Lily King

Debbie Dutridge

Dror Berel

Sheila Pressman

Inflammatory

Bowel Disease

Center: Steph Targan

Carol Landers

Dermot McGovern

Marla Dubinsky

Phillip Fleshner

Univ. of Washington

Cardiovascular

Health Research

Unit: Bruce Psaty

Josh Bis

medically refractory ulcerative colitis: gwas to translation · many ulcerative colitis patients...

Documents