the link between germline and somatic variation and

35
The link between germline and somatic variation and lifestyle risk factors in colorectal cancer Ulrike (Riki) Peters Associate Director, Public Health Science Fred Hutchinson Cancer Research Center & University of Washington

Upload: others

Post on 13-Apr-2022

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: The link between germline and somatic variation and

The link between germline and somatic variation and lifestyle risk

factors in colorectal cancer

Ulrike (Riki) PetersAssociate Director, Public Health Science

Fred Hutchinson Cancer Research Center &University of Washington

Page 2: The link between germline and somatic variation and

Genetics and Epidemiology of Colorectal Cancer Consortium (GECCO)

GECCO75+

publications

Statistical

Methods

Tumor Genome

Functional Genomics

Risk Prediction

Survival

Lifestyle & Environment

Human Genome

R01-CA059045, U01-CA137088, U19-CA148107, U01-CA164930, R01-CA176272, U01-CA185094, R01-CA201407, R01-CA206279 , R21-CA230486 X01-HG006196, X01-HG006662, X01-HG007585, X01HG009781, JUNO Therapeutics

A growing resource • 60+ studies

• CCFR and CORECT• 150,000+ participants with

genetic, clinical, epidemiologic & lifestyle data

• 3,000 participants with whole genome sequencing data

• 30,000+ patients with extended clinical and survival data

• 15,000+ patients with tumor characteristics data

• 7,000 patients with tumor sequencing data

Page 3: The link between germline and somatic variation and

Survival

Lifestyle & Environment

Human Genome

Mutated Genes

Mutational Burden

Mutated Pathways

Mutational Signatures

F. nucleatum Quantity

Existing Markers

Existing markers • BRAF• KRAS• CP Island methylation

phenotype (CIMP)• Microsatellite instability (MSI)• In 12,000 cases

Targeted tumor sequencing• 200-300 genes• Copy number changes• Fusobacterium and other

pathogens• In 7,000 cases

TumorHost

Page 4: The link between germline and somatic variation and

Smoking-CRC associations differ by molecular subtypes

3

Marker Status OR 95% CI P-trend P-difference

BRAF mut 1.20 (1.15, 1.25) 1.21E-15

wt 1.08 (1.06, 1.10) 3.76E-13 3.3E-06

KRAS mut 1.07 (1.04, 1.10) 1.11E-05

wt 1.10 (1.08, 1.13) 2.62E-16 5.5E-02

CIMP + 1.19 (1.15, 1.24) 1.31E-18

- 1.08 (1.06, 1.11) 5.48E-13 6.6E-07

MSI instable 1.16 (1.11, 1.21) 2.05E-12

stable 1.09 (1.07, 1.12) 7.83E-15 4.0E-03

0.95 1 1.05 1.1 1.15 1.2 1.25 1.3

Odds ratios

• Polytomous logistic regression • P-for difference case-only analysis

Page 5: The link between germline and somatic variation and

Smoking-CRC associations differ between groups (case-case analysis)

4

KRAS-mut

KRAS-wild

KRAS-mut

KRAS-wild

KRAS-mut

KRAS-wild

KRAS-mut

KRAS-wild

KRAS-mut

KRAS-wild

KRAS-mut

KRAS-wild

KRAS-mut

KRAS-wild

KRAS-mut

KRAS-wild

BRAF-mut

BRAF-wild

BRAF-mut

BRAF-wild

BRAF-mut

BRAF-wild

BRAF-mut

BRAF-wild

CIMP+

CIMP-

CIMP+

CIMP-

MSI-high

MSS/MSI-low

All Cases

Category P-difference

Group 1 1.4E-03

Group 10 3.5E-05

Group 8 1.7E-01

Group 5 9.1E-01

Group 2 4.3E-01

Group 6 9.6E-01

Group 9 3.1E-01

Group 7 3.0E-01

Group 3 9.1E-02

Group 4 reference

0.8 1 1.2 1.4 1.6Odds ratio

Page 6: The link between germline and somatic variation and

Molecular classification of CRC

Jass. Histopathology, 2007.

Page 7: The link between germline and somatic variation and

Colorectal cancer incidence trends by age and sex, US 1975-2014(SEER, National Program of Cancer Registries, North American Association of Central Cancer Registries)

Siegel et al, CA Cancer J Clin, 2017Year of Diagnosis

Page 8: The link between germline and somatic variation and

Tumor marker analyses by age of onset among CRC (cases-case analysis)

≥65 50-64 <50 ≥65 50-64 <50 ≥65 50-64 <50 ≥65 50-64 <500.0

0.5

1.0

1.5

Age Group

Odd

s Ra

tio

BRAF mutantKras mutantCIMP positive

P for trend = 0.22 P for trend < 0.001P for trend < 0.001

362/3377

116/2310

1105/2175

538/1165

365/2690

82/1035

MS unstable

885/4979 1877/3652 1153/4245 878/4401

412/3081

216/972

P for trend < 0.001

Ref. Ref. Ref.▪ Ref.

BRAF (mut.) KRAS (mut.) CIMP (pos.) MSI (unstable)

Yin Cao

Page 9: The link between germline and somatic variation and

Targeted Tumor Sequence of 2,100 tumors

Most Frequently Mutated Genes POLE, POLD1, MSI Status Among Hypermutated Samples (n=395, 19%)

CRC Pathway and Gene Mutations by NHM and HM Status

Zaidi, Harrison, Phipps et al. submitted

Tabitha Harrison Hassan Zaidi

Page 10: The link between germline and somatic variation and

Survival by Mutated Genes

CRC Survival by Predicted TP53 Transcriptional Activity

P-value = 3x10-5

Residual activity >5%Residual activity = 0Residual activity 0-5%

Days since diagnosis

Zaidi, Harrison, Phipps et al. submitted

Amanda Phipps

Page 11: The link between germline and somatic variation and

30 Mutational Signatures Identified

Signatures have been linked to– Aging– Smoking– UV radiation– Aristolochic acid– Aflatoxin– Mutation in APOBEC family

BRCA1/2, MMR, POLE– Treatments with alkylating agent

temozolomide– Most signatures have not been

linked to risk factors

Nik-Zainal et al. Cell 2012; Alexandrov et al. Cell Reports 2013; Alexandrov et al. Nature2013 Helleday et al. Nat Rev Genet 2014; Alexandrov et al. Curr Opin Genet Dev 2014

Ludmil Alexandrov

Page 12: The link between germline and somatic variation and

GERMLINE GENETIC RISK AND INTERACTIONS WITH ENVIRONMENTAL RISK FACTORS

Page 13: The link between germline and somatic variation and

Interactions

Colorectal Cancer Risk

Why are we interested in discovering genetic risk loci and gene-environment interactions?

• Identification of the underlying risk factors to advance understanding of disease biology

• Inform drug targets to improve treatment of disease and chemoprevention

• Modify the genetic risk that is fixed by changing the environmental exposures

• Precision prevention - Enable genetic risk prediction of diseases to guide screening and targeted interventions

Page 14: The link between germline and somatic variation and

Chromosome

Genome-wide scan of 125,000 samples identified 40 new independent CRC risk variants

Jeroen Huyghe Stephanie Bien Tabitha Harrison

Total # of genetic variants tested 16,900,000

-log

10P

Summary:• Replicate 55 previously reported signals• Discover 30 new loci with P < 5×10-8

• Discover 10 additional new conditionally independent signals with P < 5×10-8 in known and new loci

• Rare (0.6% MAF) protective variant (gene CHD1) for sporadic CRC

• Low-frequency risk variants (genes BOC, TERT, POLD3 and KLF5)

• Pathways not previously implicated by GWAS (Hedgehog signaling, Krüppel-like factors, Hippo-YAP signaling)

• Role for lncRNAs, somatic drivers and immune function

Huyghe, Bien, Harrison et al. Nat Genet 2019

Page 15: The link between germline and somatic variation and

Genetic analysis by tumor sites

Substantial effect heterogeneity between sites at established CRC risk loci- Only ½ of risk loci have shared effects

across sites and Phet>0.05- ¼ of risk loci are associated with left-

sided CRC risk, but show little evidence for association with right-sided CRC risk

Distal colon and rectal tumors share most polygenic risk

Jeroen Huyghe

Analyses based on 64,506 shared controls.

• 5 GWAS meta-analyses:

Discovery of 14 additional risk loci in GWAS stratified by primary tumor site

Left-sided CRC and right-sided CRC have to a large extent different genetic etiologies

Huyghe et al. in preparation

Tumor site N cases

Right-sided/ Proximal colon

15,706

Distal colon 14,376Rectal 16,212Colon 32,002Left-sided 30,588

Page 16: The link between germline and somatic variation and

Functionally informed GWAS

Integration of colorectal-specific super enhancers

11 novel risk loci discovered

Bien et al. in preparation

Stephanie BienPeter Scacheri

Page 17: The link between germline and somatic variation and

0

50,000,000

100,000,000

150,000,000

200,000,000

250,000,000Single variants

Indels

Based on TOPMed Freeze 5 data; Taliun et al. BioRxiv 2019

Most genetic variation is rare

Whole genome sequencing of 65,000 samples, total 470M variants detected

Page 18: The link between germline and somatic variation and

Integration of functional genomics to improve discoveries for GxE interactions

RNA sequencing gene expression

DNA accessibility• ATAC-Seq

Machine learning

R01-CA201407, MPI: Casey, Gauderman, Le Marchand and Peters

GxE100,000

cases and controls

with GWAS & E

Genotyping500biopsies

50organoids

Page 19: The link between germline and somatic variation and

Harmonized epidemiological data across studies

Odds ratios adjusted for age, sex and study

Harmonization of individual level data ongoing for 60+ studies across CCFR, CORECT and GECCO at the coordinating center using a consistent pipeline

Page 20: The link between germline and somatic variation and

Genome-wide GxE findings for colorectal cancerFindings with p-value < 5x10-08 in up to 27k participants from CCFR and GECCO

Epi Variable Chr / gene P-value for GxE Publication/StatusProcess. Meat 10p14/GATA3 8.7E-09

Figueiredo et al. PLoS Genet 2014Red meat NoneFruits NoneVegetables NoneFiber NoneNSAIDS 12p12.3/HGST1 4.6E-09 Nan et al. JAMA 2015

15q25.2/IL16 8.2E-09HRT 20q13.2/CY24A1 4.8E-09 Garcia d.A., Rudolph Br J Can 2016Calcium None Du et al. CEBP 2015Alcohol 9q22.32/HIATL1 1.8E-08

Gong et al. PLoS Genet 2016Smoking NoneFolate 1p22.1/ABCA4 3.0E-08 Du et al. (in preparation)

9p21.3/FOCAD 7.9E-09BMI 118q21.1/SMAD7 Cocktail Campbell et al. (in preparation)Height NonePhysical activity None Joshi et al. (in preparation)Diabetes None Murphy (expanding #s)

Page 21: The link between germline and somatic variation and

NIH Director’s Blog highlighted our NSAIDsxG discovery (Nan et al. JAMA 2015) as an example for precision medicine

GxE enables precision medicine

Page 22: The link between germline and somatic variation and

Utilizing or discoveries for risk prediction

Page 23: The link between germline and somatic variation and

Survival depends on early detection

STAGEI II III IV

5-year survival 75%

5-year survival 30-70%

5-year survival 6%

STAGE STAGE STAGE

Page 24: The link between germline and somatic variation and

Screening is effective in early detection

• Available screening tools endoscopy, blood in stool test

• About 1/3 of eligible people do not receive colorectal cancer screeningSerious side effects, inconvenience, costs, others

• Factors influencing screening decision in general population Age, family history

Page 25: The link between germline and somatic variation and

Average risk vs. personal risk

Average risk to develop colorectal cancer in the next 10 years is

close to 1% for an adult age 50

Risk of colorectal cancer

highlow

Freq

uenc

y in

the

popu

latio

n

Page 26: The link between germline and somatic variation and

Build comprehensive risk prediction model

• Improve screening efficiency through risk stratified screening

• Identify high risk group for targeted preventive intervention– Lifestyle (diet, smoking, etc)– NSAIDs

Page 27: The link between germline and somatic variation and

Recommended age to start colorectal cancer screening based on genetic and environmental risk factors profile

Current recommended starting age for screening

Risk Score (%) Risk Score (%)

Jeon, Hsu et al. Gastroenterol 2016 and 2018

Risk threshold set to 0.89% (average 10-year risk for an individual aged 50 years without prior endoscopy)

Li Hsu Jihyoun Jeon

Page 28: The link between germline and somatic variation and

Genome-wide risk prediction

• Genome-wide risk prediction model:• Pruning SNPs and then build (non-linear)

prediction model using machine learning algorithms (Prune + ML)

• Using the entire genome to build prediction model (LDPred)

Model # Variants AUC

Known Loci 120 0.63

Prune + ML 13,000 0.64

LDPred 3.6million 0.68

Chad He Jessica Minnier Minta Thomas

90

10%

5%

Dise

ases

Pro

babi

lity

40 50 60 70 80Age

90%

25-75%

<10%

Page 29: The link between germline and somatic variation and

Net discounted costs compare uniform and risk-stratified screening with equal effectiveness

Risk-stratified screening based on risk prediction algorithms that has an AUC=0.70 would reduce resources needed for screening by >$250,000 per 1,000 individuals screened at similar effectiveness as current screening recommendations (Costs for risk assessment, i.e. genetic testing and questionnaire data were not taken into account)

Blue = current screening guidelines; red = risk-stratified screening

Preliminary data based on the MISCAN-Colon model

Iris Lansdorp-Vogelaar

Elleke Peterse

Ann Zauber

Josh Roth

Page 30: The link between germline and somatic variation and

Evaluation of risk-stratified screening in Kaiser Permanente Northern CA cohort and Estonia Biobank

Study population Total

All successfully genotyped 102,979

Colorectal cancer 744

Colorectal adenoma 11,830

Advanced adenoma 4,614

Hyperplastic polyp 3,589

Had colonoscopy 32,733

Had hemoccult test 69,965

R01-CA206279; MPI: Corley, Hayes, Peters

Doug Corley

Peeter Padrik

Tonis Tasa

Estonian Biobank • 150,000 participants in

Biobank – ~12% of entire population– 700 Colorectal cancer cases– GWAS– Nationwide EHR system

Kaiser Permanente cohort

Page 31: The link between germline and somatic variation and

Genic risk score (based on known GWAS loci) is particularly predictive for early onset cases without a family history

Archambault et al. under review

Early Onset CRCWe have >5000 CRC cases diagnosed <50 years of age

Alexi Archambault Richard Hayes

Page 32: The link between germline and somatic variation and

Moving towards risk stratified screening in a prospective trial

60,000 participants age 45-60

Risk stratified screening

Current screening recommendations

Follow up for advanced adenoma & cancer

Page 33: The link between germline and somatic variation and

Commercialization of polygenic risk scores

• Myriad has included polygenic risk score in genetic testing since 2017

• Other companies are following, such as Ambry Genetics, Color Genetics,…

• Polygenic risk scores as one of the top 10 Breakthrough Technologies in 2018 by MIT Technology Review

Page 34: The link between germline and somatic variation and

Summary

• Some environmental and clinical factors show strong associations with mutational profiles

• About 120 common genetic risk variants have been identified for colorectal cancer

• 8 gene-environment interactions have been detected• A large fraction of the common and rare genetic variation remains to be

discovered• Using genetic + environmental data to inform screening holds promise and

needs to be evaluated in a prospective setting

Page 35: The link between germline and somatic variation and

GECCO meeting, Seattle 2019THANK YOU!