Cancer Biomarker:Cancer Biomarker:Single Nucleotide PolymorphismSingle Nucleotide Polymorphism

Dong Hwan Kim, MD/PhD

Samsung Medical CenterSungkyunkwan University School of Medicine

Seoul, Korea

• Basic concepts of SNPs

• Applications of SNPs into cancer research

• Sampling issue in SNP study

• Methodology & Technology in SNP study

• SNP selection in SNP study

• Biostatistical analysis in SNP study

• Validation in SNP study

ContentsContents

• Basic concepts of SNPs

• Applications of SNPs into cancer research

• Sampling issue in SNP study

• Methodology & Technology in SNP study

• SNP selection in SNP study

• Biostatistical analysis in SNP study

• Validation in SNP study

Single Nucleotide PolymorphismSingle Nucleotide Polymorphism

Basic ConceptBasic Concept

What is SNP?What is SNP?((SSingle ingle NNucleotide ucleotide PPolymorphism)olymorphism)

• Single base pair change in the genomic DNA allele.

• not a somatic mutation,

• but consitutional variation (=germline)

• Single base pair change in the genomic DNA allele.

• not a somatic mutation,

• but consitutional variation (=germline)

Frequency of SNPFrequency of SNP1/1,000 nucleotides1/1,000 nucleotides→ 10 million SNPs→ 10 million SNPs

Frequency of SNPFrequency of SNP1/1,000 nucleotides1/1,000 nucleotides→ 10 million SNPs→ 10 million SNPs

cc--/s/s--/r/r--SNPsSNPs(coding / silent / regulatory)(coding / silent / regulatory)

Amino acid changePromoter

Splicing variant

Protein expressionProtein expressionchangechange

Single Nucleotide PolymorphismSingle Nucleotide Polymorphism

Applications of SNPsinto cancer research

Single Nucleotide PolymorphismSingle Nucleotide Polymorphism

Applications of SNPsinto cancer research

Phenotypes

Bioinformatics-Genotypic model

-Allele model-Haplotype-Tag SNP

Bioinformatics

Genotyping

-Single gene-Candidate pathway

-Genome wide

Genotypes

+

Test set

Phenotype record-Susceptibility

-Response-Drug toxicity

-Survival

DNA sample collection (tumor DNA vs germline DNA)

Test set

Validation set

Validation set

Construction of Predictive Model

Association StudyAssociation Study

1. Purpose: What SNP of what gene is associated with increasing susceptibility to certain type of cancer?

2. Target population: certain type of cancer -> Sex/Age matched cohort (Case and Control)

3. Hypothesis based or Discovery oriented ?1) Hypothesis based -> Candidate gene or pathway approach2) Discovery oriented -> Genome wide approach

4. Candidate gene / pathway approacheg. Apoptosis -> caspase 3 / 8 / 9 genes, BCL2 gene etc.-> frequency of these gene SNPs between case vs. control

5. Genome wide association study (GWAS)eg. Whole genome 1 million SNPs-> detect significant signals-> validate -> define main effector SNP

1. Purpose: What SNP of what gene is associated with increasing susceptibility to certain type of cancer?

2. Target population: certain type of cancer -> Sex/Age matched cohort (Case and Control)

3. Hypothesis based or Discovery oriented ?1) Hypothesis based -> Candidate gene or pathway approach2) Discovery oriented -> Genome wide approach

4. Candidate gene / pathway approacheg. Apoptosis -> caspase 3 / 8 / 9 genes, BCL2 gene etc.-> frequency of these gene SNPs between case vs. control

5. Genome wide association study (GWAS)eg. Whole genome 1 million SNPs-> detect significant signals-> validate -> define main effector SNP

Pharmacogenetic StudyPharmacogenetic Study

1. Purpose: What SNP of what gene is associated with response or drug toxicity to anticancer therapy in certain type of cancer?

2. Target population: specific type of cancer, specific type of therapy with specific schedule: New agent + conventional tx?

3. Hypothesis based or Discovery oriented ?1) Hypothesis based -> Candidate gene or pathway approach2) Discovery oriented -> Genome wide approach

4. Candidate gene / pathway approacheg. UGT1A1 gene SNPs or FCGR3A gene SNPs-> Response % or toxicity % according to genotypes

5. Genome wide approacheg. Whole genome 1 million SNPs-> detect significant signals-> validate -> define main effector SNP

1. Purpose: What SNP of what gene is associated with response or drug toxicity to anticancer therapy in certain type of cancer?

2. Target population: specific type of cancer, specific type of therapy with specific schedule: New agent + conventional tx?

3. Hypothesis based or Discovery oriented ?1) Hypothesis based -> Candidate gene or pathway approach2) Discovery oriented -> Genome wide approach

4. Candidate gene / pathway approacheg. UGT1A1 gene SNPs or FCGR3A gene SNPs-> Response % or toxicity % according to genotypes

5. Genome wide approacheg. Whole genome 1 million SNPs-> detect significant signals-> validate -> define main effector SNP

• Single gene based approach– IL10 gene– VEGF gene– MDR1 gene

• Candidate pathway based approach - apoptosis

• Genome wide approach – whole genome

• Single gene based approach– IL10 gene– VEGF gene– MDR1 gene

• Candidate pathway based approach - apoptosis

• Genome wide approach – whole genome

• Advantages– 가설에근거한연구로결과해석이용이함– 샘플수가적어도가능– 분석이단순함 / 저비용– validation이용이함

• Disadvantages– 현재까지알려진 SNP정보에의존하므로 true signal을

miss할가능성존재– polygenic trait가존재하는경우에는검출이어려움– 유전자간의상호작용고려않음

Single gene approachSingle gene approach

• Advantages– 가설에근거한연구로결과해석이용이함– 샘플수가적어도가능– 분석이단순함 / 저비용– validation이용이함

• Disadvantages– 현재까지알려진 SNP정보에의존하므로 true signal을

miss할가능성존재– polygenic trait가존재하는경우에는검출이어려움– 유전자간의상호작용고려않음

• Hypothesis based single gene approach– FCGR3A gene SNP affect response and treatment outcome of pts with

lymphoma receiving R-CHOP chemotherapy– Case : lymphoma, receiving R-CHOP (n=113) or CHOP (n=85)– Control : not needed– Korean pts

• Candidate single gene:– 2 SNP markers involved in the FCGR3A and FCGR2A gene polymorphisms

• Results: pharmacogenetic study– FCGR3A VV allele associated with better and faster response to R-CHOP

therapy

Example 1:Example 1: Rituximab and FCGR3A SNPsRituximab and FCGR3A SNPs• Hypothesis based single gene approach

– FCGR3A gene SNP affect response and treatment outcome of pts with lymphoma receiving R-CHOP chemotherapy

– Case : lymphoma, receiving R-CHOP (n=113) or CHOP (n=85)– Control : not needed– Korean pts

• Candidate single gene:– 2 SNP markers involved in the FCGR3A and FCGR2A gene polymorphisms

• Results: pharmacogenetic study– FCGR3A VV allele associated with better and faster response to R-CHOP

therapy

Kim DH, et al. Blood 2006

R-CHOP group CHOP group

• Hypothesis based single gene approach– UGT1A SNP affect toxicity and treatment outcome of pts with lung cancer– Case : NSCLC, receiving Irinotecan (80mg/m2) on D1 / D8 and cisplatin

60mg/m2 on D1– Control : not needed– Korean 81 pts

• Candidate single gene: – 8 SNP markers involved in the UGT1A polymorphism– UGT1A9*22 (rs3832043)– UGT1A7*1,*2,*3,*4 N129K (rs17868323) / 1A7*1,*2,*3,*4 R131R/Q/K

(rs17863778)1A7*1,*2,*3,*4 R131R/Q/K (rs17868324) / 1A7*1,*2,*3,*4 W208R

(rs11692021)– UGT1A1*60 -3279T>G (rs4124874) / -53~-42(TA)6/(TA)7 (UGT1A1*28)

(rs8175347)1A1*6 211G>A (G71R) (rs4148323)

• Results: pharmacogenetic study– UGT1A1*6 and UGT1A9*22 genotypes may be important for SN-38

glucuronidation and associate with irinotecan-related severe toxicity (neutropenia and diarrhea).

Example 2:Example 2: Irinotecan and UGT1A SNPsIrinotecan and UGT1A SNPs• Hypothesis based single gene approach

– UGT1A SNP affect toxicity and treatment outcome of pts with lung cancer– Case : NSCLC, receiving Irinotecan (80mg/m2) on D1 / D8 and cisplatin

60mg/m2 on D1– Control : not needed– Korean 81 pts

• Candidate single gene: – 8 SNP markers involved in the UGT1A polymorphism– UGT1A9*22 (rs3832043)– UGT1A7*1,*2,*3,*4 N129K (rs17868323) / 1A7*1,*2,*3,*4 R131R/Q/K

(rs17863778)1A7*1,*2,*3,*4 R131R/Q/K (rs17868324) / 1A7*1,*2,*3,*4 W208R

(rs11692021)– UGT1A1*60 -3279T>G (rs4124874) / -53~-42(TA)6/(TA)7 (UGT1A1*28)

(rs8175347)1A1*6 211G>A (G71R) (rs4148323)

• Results: pharmacogenetic study– UGT1A1*6 and UGT1A9*22 genotypes may be important for SN-38

glucuronidation and associate with irinotecan-related severe toxicity (neutropenia and diarrhea).

Han JY, et al. J Clin Oncol 2006

• Advantages– single gene approach보다는검출력향상– polygenic trait가존재하는경우유전자간의상호작용고려가능

• Disadvantages– 현재까지의정보에의존함– 많은수의샘플사이즈필요 (≥ 100)– 각각의유전자가동일한가중치를가진다는가설을전제로함

– 유전자간상호작용을생물학적으로검증하기힘듦– Validation이힘듦

Candidate pathway approachCandidate pathway approach

• Advantages– single gene approach보다는검출력향상– polygenic trait가존재하는경우유전자간의상호작용고려가능

• Disadvantages– 현재까지의정보에의존함– 많은수의샘플사이즈필요 (≥ 100)– 각각의유전자가동일한가중치를가진다는가설을전제로함

– 유전자간상호작용을생물학적으로검증하기힘듦– Validation이힘듦

• Hypothesis based association study– susceptibility to CML associated with apoptosis pathway– Case : CML (n=170) / Control : healthy normal (n=182)– White population

• Candidate pathways:– 80 SNP markers involved in the pathways of

apoptosis (n=30), angiogenesis (n=7), myeloid cell growth (n=14), xenobiotic metabolism (n=13), WT1 signaling (n=7), interferon signaling (n=4) and others (n=5)

• Results: internal validation– only BCL2 (rs1801018) was significantly associated with the

susceptibility to CML (p=0.05; adjusted-OR 2.16, [1.00-4.68]).– a greater number of risk alleles from BCL2 SNP correlated to increasing

risk of CML for 3-4 risk alleles vs. 0-1 risk alleles.

Example 3:Example 3: CML and BCL2 SNPsCML and BCL2 SNPs• Hypothesis based association study

– susceptibility to CML associated with apoptosis pathway– Case : CML (n=170) / Control : healthy normal (n=182)– White population

• Candidate pathways:– 80 SNP markers involved in the pathways of

apoptosis (n=30), angiogenesis (n=7), myeloid cell growth (n=14), xenobiotic metabolism (n=13), WT1 signaling (n=7), interferon signaling (n=4) and others (n=5)

• Results: internal validation– only BCL2 (rs1801018) was significantly associated with the

susceptibility to CML (p=0.05; adjusted-OR 2.16, [1.00-4.68]).– a greater number of risk alleles from BCL2 SNP correlated to increasing

risk of CML for 3-4 risk alleles vs. 0-1 risk alleles.

Kim DH, et al. Blood 2009 [epub]

• Advantages– 현재의정보에근거하지않고, novel marker를검출할수있고, 가설을생성할수있음

– 유전자간의상호작용에대한 non-biased global assess가능

• Disadvantages– 위양성결과가매우많음– 매우많은샘플사이즈를요구 (≥ 500~1,000)– 통계적처리문제 (Bioinformatics, supercomputer)– 고비용 (300-500 USD / sample)– 환율

Genome wide approachGenome wide approach

• Advantages– 현재의정보에근거하지않고, novel marker를검출할수있고, 가설을생성할수있음

– 유전자간의상호작용에대한 non-biased global assess가능

• Disadvantages– 위양성결과가매우많음– 매우많은샘플사이즈를요구 (≥ 500~1,000)– 통계적처리문제 (Bioinformatics, supercomputer)– 고비용 (300-500 USD / sample)– 환율

• Genome wide study – prognosis & pharmacogenetics– Treatment response to ALL in Genome wide scale

– Case : St Jude (n=318) / COG (n=169), treated with uniform protocol

• Genome wide association analysis for MRD:– Discovery cohort (St Jude), Validation cohort (COG group)

– 588,920 SNPs --à476,796 germline SNPs --à 102 significant SNPs

21 SNPs a/w relapse

21 SNPs a/w antileukemic drug disposition

-> 63 SNPs a/w early response, relapse or drug disposition

• Independent Validation using 2 independent cohorts

Example 4:Example 4: Genome wide studyGenome wide study

• Genome wide study – prognosis & pharmacogenetics– Treatment response to ALL in Genome wide scale

– Case : St Jude (n=318) / COG (n=169), treated with uniform protocol

• Genome wide association analysis for MRD:– Discovery cohort (St Jude), Validation cohort (COG group)

– 588,920 SNPs --à476,796 germline SNPs --à 102 significant SNPs

21 SNPs a/w relapse

21 SNPs a/w antileukemic drug disposition

-> 63 SNPs a/w early response, relapse or drug disposition

• Independent Validation using 2 independent cohorts

Yang JJ, et al. JAMA 2009

Single gene vs. Candidate pathway vs. Genome wideSingle gene vs. Candidate pathway vs. Genome wide

장점 단점

Single gene • 가설에근거한연구로결과해석이용이함

• 샘플수가적어도가능• 분석이단순함• 저비용• validation이용이함

• 현재까지알려진 SNP정보에의존하므로진정한 signal을 miss할가능성존재함

• polygenic trait가존재하는경우에는검출이어려움

• 유전자간의상호작용고려않음

Candidate Pathway based

• single gene approach보다는검출력향상

• polygenic trait가존재하는경우유전자간의상호작용고려가능

• 현재까지의정보에의존함• 많은수의샘플사이즈필요• 각각의유전자가동일한가중치를가진다는가설을전제로함

• 유전자간상호작용을 생물학적으로검증하기힘듦

• Validation이힘듦

Candidate Pathway based

• single gene approach보다는검출력향상

• polygenic trait가존재하는경우유전자간의상호작용고려가능

• 현재까지의정보에의존함• 많은수의샘플사이즈필요• 각각의유전자가동일한가중치를가진다는가설을전제로함

• 유전자간상호작용을 생물학적으로검증하기힘듦

• Validation이힘듦Genome-wide • 현재의정보에근거하지않고,

novel marker를 검출할수있고, 가설을생성할수있음

• 유전자간의상호작용에대한non-biased global assess가능

• 위양성결과가매우많음• 매우많은샘플사이즈를요구• 통계적처리문제• 고비용

Single Nucleotide PolymorphismSingle Nucleotide Polymorphism

Sampling issuesSampling issues

• Germline DNAs– from peripheral blood, oral swab, fresh frozen or formalin embedded

normal tissue– eg) susceptibility analysis to certain type of cancer– eg) chemotherapy toxicity (diarrhea, skin rash, neutropenia)

• Tumor DNAs– from cryopreserved or formalin embedded tumor tissue– even can isolate circulating tumor DNAs from plasma or serum– eg) unrevealed amplification, deletion, chromosomal abnormalities,

changes of copy numbers in tumor cells – Discordance of germline and tumor genotypes

• Selection of sample type is very important even at the initial development of project (study design).

• Depending on the purpose of the study, study design will be determined, then the type of samples also will be decided.

Germline vs. tumor DNAsGermline vs. tumor DNAs

• Germline DNAs– from peripheral blood, oral swab, fresh frozen or formalin embedded

normal tissue– eg) susceptibility analysis to certain type of cancer– eg) chemotherapy toxicity (diarrhea, skin rash, neutropenia)

• Tumor DNAs– from cryopreserved or formalin embedded tumor tissue– even can isolate circulating tumor DNAs from plasma or serum– eg) unrevealed amplification, deletion, chromosomal abnormalities,

changes of copy numbers in tumor cells – Discordance of germline and tumor genotypes

• Selection of sample type is very important even at the initial development of project (study design).

• Depending on the purpose of the study, study design will be determined, then the type of samples also will be decided.

• 전이성위암환자에서새로운신약인 kutnezumomab과capecitabine+cisplatin 항암치료에대한부작용인 grade 3,4 diarrhea와 FCGR3A유전자의 SNP에대한관계를보고자한다.

• 전체환자들중 1/3은항암치료를종결하였고, 1/3은항암치료중, 1/3은항암치료를계획중이다. 종양조직은 1/3에서만있고, 사망자는 2명발생하였고, 이환자의경우종양조직에대한파라핀조직슬라이드만보관되어있었다.

• 연구를위한샘플은? 샘플링시점은?

• Peripheral blood sampling (5cc EDTA)

• Use of paraffin tissue – in debate

Example: Sampling issueExample: Sampling issue

• 전이성위암환자에서새로운신약인 kutnezumomab과capecitabine+cisplatin 항암치료에대한부작용인 grade 3,4 diarrhea와 FCGR3A유전자의 SNP에대한관계를보고자한다.

• 전체환자들중 1/3은항암치료를종결하였고, 1/3은항암치료중, 1/3은항암치료를계획중이다. 종양조직은 1/3에서만있고, 사망자는 2명발생하였고, 이환자의경우종양조직에대한파라핀조직슬라이드만보관되어있었다.

• 연구를위한샘플은? 샘플링시점은?

• Peripheral blood sampling (5cc EDTA)

• Use of paraffin tissue – in debate

Single Nucleotide PolymorphismSingle Nucleotide Polymorphism

Methodology & TechnologyMethodology & Technology

Principles of genotyping1. Allele discrimination reactions– DNA polymerases

• Single-base extension (SBE)• Allele-specific extension and allele-specific PCR (AS-PCR)• Synchronized DNA synthesis - pyrosequencing• Structure specific cleavage - the invader assay• The 5’ exonuclease activity - the Taqman assay

– DNA ligases– Hybridization

1. Allele discrimination reactions– DNA polymerases

• Single-base extension (SBE)• Allele-specific extension and allele-specific PCR (AS-PCR)• Synchronized DNA synthesis - pyrosequencing• Structure specific cleavage - the invader assay• The 5’ exonuclease activity - the Taqman assay

– DNA ligases– Hybridization

2. Detection and identification of allele-specific products– Fluorescence resonance energy transfer (FRET)– Fluorescence polarization (FP)– Mass spectrometry– Microarray as supporter for genotyping– Fluorescence coded microbeads as supporter– Seperation of products by electrophoresis

Comparisons of different technologiesComparisons of different technologies종류 설명 장점 단점

Chip based technology

• Affymetrix or illumina platform

• 250K-1M의 SNP을동시에검사

• Discovery oriented research에적합

•많은수의 SNP을한꺼번에비교적적은양의샘플로검사

•전체 SNP을검사하는것을감안하면효율적인비용으로검사가가능

•비교적높은 call rate

•한샘플당가격이 150-500 USD정도로비싸서, 많은샘플사이즈의연구에서는비용이많이든다.

•다량의데이터를분석할bioinformatic support가 전제.

•내가원하는 SNP이 chip에포함되어있지않는경우가발생가능

High throughput technology (or mass ARRAY)

• Sequenom의 MALDI-TOF based platform

• Hypothesis based research에적합

• Candidate pathway or gene based apporoach에적합

•전통적인방법에비해많은수의샘플을짧은시간안에처리가능

• Multiplex PCR이가능하여비용절감 (0.2-0.3 USD/샘플/genotype)

•비교적높은 call rate

•많은샘플의 genotype결과를판독할 machine을 구매할필요가있고이의가격이비싸다.

•적은수의샘플인경우에는저비용의장점이소실된다.

•플랫폼에따라검출이불가능한 SNP이존재가능

High throughput technology (or mass ARRAY)

• Sequenom의 MALDI-TOF based platform

• Hypothesis based research에적합

• Candidate pathway or gene based apporoach에적합

•전통적인방법에비해많은수의샘플을짧은시간안에처리가능

• Multiplex PCR이가능하여비용절감 (0.2-0.3 USD/샘플/genotype)

•비교적높은 call rate

•많은샘플의 genotype결과를판독할 machine을 구매할필요가있고이의가격이비싸다.

•적은수의샘플인경우에는저비용의장점이소실된다.

•플랫폼에따라검출이불가능한 SNP이존재가능

Conventional technology

• PCR/RFLP• Pyrosequencing• Allele-specific PCR• DHPLC• Taqman genotyping

•비싼장비구입이필요하지않아가격이싸다.

•소수의샘플을대상으로소수의 SNP만을보고자할경우선택

• Labor intensive•따라서많은수의샘플처리나, 많은수의 SNP검사에는적절하지않음

• Call rate가비교적낮다

High throughput technologyHigh throughput technology

Genome wide

GenomeGenome--wide SNP arraywide SNP array

Affymetrix illumina

Affymetrix vs. illumina

Affymetrix SNP 6.0 illumina 1MTotal # Genetic Markers ~1.8 million ~1.0 millionNo. of SNPs ~906K ~1,050KSNP selection Random Tag SNPRacial preference Non CaucasianNon CaucasianCall rates 99.8% 99.66%No. CNV probes ~946K ~22KMedian Marker Spacing 680 bases 1,700 basesNo. of samples for $1M 4,000 2,000DNA required 500 ng 400 ngPower at GRR=1.5, AF=0.1, Prevalence=0.1

0.90 0.25

• BCL2 gene expression profile을보다가 25개의lymphoma cell line마다 BCL2의 genotype에따른 BCL2의 mRNA RQ-PCR의차이가있나를보고자한다.– Conventional methods (25 cases, single gene SNPs)

• 급성백혈병환자 110명을대상으로약재대사및수송기전을알고있는 Methotrexate의부작용인구강점막염의빈도차이를결정하는 SNP을알고자한다.– High-troughput tech. (110 cases, candidate pathway -> MTX

metabolism & transport full cover)

Example 1: MethodologyExample 1: Methodology

• BCL2 gene expression profile을보다가 25개의lymphoma cell line마다 BCL2의 genotype에따른 BCL2의 mRNA RQ-PCR의차이가있나를보고자한다.– Conventional methods (25 cases, single gene SNPs)

• 급성백혈병환자 110명을대상으로약재대사및수송기전을알고있는 Methotrexate의부작용인구강점막염의빈도차이를결정하는 SNP을알고자한다.– High-troughput tech. (110 cases, candidate pathway -> MTX

metabolism & transport full cover)

• 새로운신약인 Popopotinib의부작용으로알려진 skin

rash의기전을밝히고자한다. 전체환자수는 200명인데

이중 50명에서 skin rash가발생하였다.

– 제안된기전이현재까지아무것도없다.

– Genome wide SNP array

– Issue to be resolved

• 200명 x 500 USD/cases = 100,000 USD (funding issue)

• 1,000,000 genotype에 대해분석해야함 (bioinformatic support)

Example 2: MethodologyExample 2: Methodology

• 새로운신약인 Popopotinib의부작용으로알려진 skin

rash의기전을밝히고자한다. 전체환자수는 200명인데

이중 50명에서 skin rash가발생하였다.

– 제안된기전이현재까지아무것도없다.

– Genome wide SNP array

– Issue to be resolved

• 200명 x 500 USD/cases = 100,000 USD (funding issue)

• 1,000,000 genotype에 대해분석해야함 (bioinformatic support)

Possible reasons for lack of success of genetic studies

1. Study design– Poor matching of cases and controls in association studies

2. Sample size– Too small to detect genes of likely effect size for complex traits

3. Phenotyping quality– Many disease phenotypes are of low or unknown reliability, leading to

extreme difficulty in replication

4. Genotyping quality– Genotyping error rates can be > 3-5% in practice

5. Relationship errors– Mistaken inference about family structure in linkage analysis

6. Publication bias– The large number of single reports of marginally significant findings

on ‘candidate genes’ lends great difficulty to interpretation & meta-analysis

7. Validation issues

1. Study design– Poor matching of cases and controls in association studies

2. Sample size– Too small to detect genes of likely effect size for complex traits

3. Phenotyping quality– Many disease phenotypes are of low or unknown reliability, leading to

extreme difficulty in replication

4. Genotyping quality– Genotyping error rates can be > 3-5% in practice

5. Relationship errors– Mistaken inference about family structure in linkage analysis

6. Publication bias– The large number of single reports of marginally significant findings

on ‘candidate genes’ lends great difficulty to interpretation & meta-analysis

7. Validation issues

Single Nucleotide PolymorphismSingle Nucleotide Polymorphism

SNP selectionSNP selection

• Discovery oriented approach– Genome wide -> “-omics” / DMET chip

• Hypothesis oriented approach– Candidate gene pathway approach

• How to select genes in multiple candidate pathways?– Pharmacogenetics: www.pharmgkb.org

– Other molecular pathways• reactome.org

• snp3d.org

• www.genome.ad.jp ”KEGG”

SNP Approach

• Discovery oriented approach– Genome wide -> “-omics” / DMET chip

• Hypothesis oriented approach– Candidate gene pathway approach

• How to select genes in multiple candidate pathways?– Pharmacogenetics: www.pharmgkb.org

– Other molecular pathways• reactome.org

• snp3d.org

• www.genome.ad.jp ”KEGG”

Gene-gene interaction

www.snp3d.org

Drug metabolism pathway

www.pharmgkb.org

Candidate pathway search

www.genome.ad.jp

• Literature review

• SNP500 site (http://snp500cancer.nci.nih.gov/)

• FastSNP site(http://fastsnp.ibms.sinica.edu.tw/pages/input_CandidateGeneSearch.jsp)

• NCBI entrez site(http://www.ncbi.nlm.nih.gov/SNP/)

SNP selection

• Literature review

• SNP500 site (http://snp500cancer.nci.nih.gov/)

• FastSNP site(http://fastsnp.ibms.sinica.edu.tw/pages/input_CandidateGeneSearch.jsp)

• NCBI entrez site(http://www.ncbi.nlm.nih.gov/SNP/)

NCBI SNP site

FastSNP

SNP500 site

• DMET Plus contains 1,936 markers in 225 genes.

• DMET Core – 169 markers– Features markers for all FDA validated genes– Based on ADME Core as defined by

PharmaADME group www.pharmaADME.org• Covers 100% of the genes and >90% of the

markers

• DMET Extended– Recommended by Key Opinion Leaders in

Industry and Academia (KOL Recommended) –147 markers

• Includes FDA Probable Valid biomarkers– Supplemental ADME – 1,621 markers

– Additional markers in drug metabolism and transporter genes

DMET (Drug METabolism) PlusTM

• DMET Plus contains 1,936 markers in 225 genes.

• DMET Core – 169 markers– Features markers for all FDA validated genes– Based on ADME Core as defined by

PharmaADME group www.pharmaADME.org• Covers 100% of the genes and >90% of the

markers

• DMET Extended– Recommended by Key Opinion Leaders in

Industry and Academia (KOL Recommended) –147 markers

• Includes FDA Probable Valid biomarkers– Supplemental ADME – 1,621 markers

– Additional markers in drug metabolism and transporter genes

Phase I Phase II

414 markers658 markers

CYP2D6CYP2C9CYP1A1CYP1B1CYP2C19DPYD

NAT1NAT2GSTM1GSTT1UGT1A1SULT1A

DMET PlusTM

AHRATP7ANR1┃2PPARD

Transpoter

552 markersModifier

137 markers

Related Gene

175 markers

ABCB1ABCC2SLC15A2SLCO2B1

ABP1COMT…VKORC1

Single Nucleotide PolymorphismSingle Nucleotide Polymorphism

BiostatisticsBiostatisticsBioinformaticsBioinformatics

Single Nucleotide PolymorphismSingle Nucleotide Polymorphism

BiostatisticsBiostatisticsBioinformaticsBioinformatics

Very important issue!However,

It is not my area.

Single Nucleotide PolymorphismSingle Nucleotide Polymorphism

ValidationValidation

• Internal validation• Permutation - bootstrap method• may justify your study, but does not guarantee it is

enough.• External validation

• Learning set vs. validation set• If possible, build up from different center

(independent cohort / collaboration)• Predictive model issue

• In debate• May or may not another validation set to validate

predictive model

Step for SNP studyStep for SNP study

• Internal validation• Permutation - bootstrap method• may justify your study, but does not guarantee it is

enough.• External validation

• Learning set vs. validation set• If possible, build up from different center

(independent cohort / collaboration)• Predictive model issue

• In debate• May or may not another validation set to validate

predictive model

Phenotypes

Bioinformatics-Genotypic model

-Allele model-Haplotype-Tag SNP

Bioinformatics

Genotyping

-Single gene-Candidate pathway

-Genome wide

Genotypes

+

Test set

Phenotype record-Susceptibility

-Response-Drug toxicity

-Survival

DNA sample collection (tumor DNA vs germline DNA)

Test set

Validation set

Validation set

Construction of Predictive Model

Single Nucleotide PolymorphismSingle Nucleotide Polymorphism

Summary of SNP studiesSummary of SNP studies

Conclusion

• Basic concepts of SNPs • Association study / pharmacogenomics• Sampling issues (tumor vs germline DNAs)• Genome-wide / high throughput / conventional• SNP selection• Validation

• Basic concepts of SNPs • Association study / pharmacogenomics• Sampling issues (tumor vs germline DNAs)• Genome-wide / high throughput / conventional• SNP selection• Validation

The Genomics gold rush JAMA 2007;298:218-221