Coursera 코스 "유전체학과 정밀의학"

Week 1: Human Genome Strcture, Function, and Variation #

The structure of the human genome and how things work #

Human genome

How is gene expression regulated?

• Protein-DNA in Promoter
• RNA-DNA in Promoter
• Epigenetic in Promoter

Human genetic variation - structural variation #

Big changes of chromosome

• Translocation: 1/600 newborn
• CNVs: deletions or duplications
  • 68% of CNVs overlap with genes
  • recessive CNVs
  • Size and location are pathogenic

Human genetic variation - single nucleotide variants #

Mutation

• Random mutations arise natually during cell division
• gemline and somatic
• 50 to 100 de novo mutations in average newborn

SNVs

• Ancestral and minor allele
• Allele frequency, rare(<0.5%, 60M no of SNVs in human pupulation), low frequency(0.5-5%, 20M), common variants(>5%, 10M)
• Average person has ~4M DNA variants
• Minor allele frequency of ethnic group tells the human migration
• Minor > 50% means that allele is beneficial in that's environment

Consquences of single nucleotide variants in genes #

Types of SNVs

• Synnymous (slient)
• Non-synomous (missense)
• Premature stop (nonsense)
• Frameshift

Average person ~350 loss of function SNVs

Architecture of human genetic variation #

Haplotype

• Only partial combination of SNVs
• Recombination breaks up haplotype
• Linkage disequilibrium (LD)
• LD allows prediction of alleles

Week 2: Applying Genomics to Medicine #

Background and Mendelian inheritance of disease #

Meldelian disorder

• present in ~2-3% of all newborn
• OMIM (Online Mendelian Inheritance in Man)

3 major inheritance patterns

1. Autosomal dominant: 50%
2. Autosomal recessive
3. X-linked

It's not so simple

• New mutation
• Mosaicism - The exception to every cell having the same DNA
• Decreased Penetrance - Disease genotype without phenotype
• Disease in Carriers of Recessive Disorders - Not always "silent"

Increased Disease Risk in Carriers

• Gaucher Disease
• Ataxia Telangectasia
• Cystic fibrosis
• Sickle cell trait

Genetic testing for Mendelian diseases #

Matching the test to the variant

Newborn Screening #

A latent or early symptomatic stage exists during which intervention imporves outcomes.

Successful newborn screening requires:

1. Acquiring the samples and performing the test
2. Having a system for notifying families whose infants test positive
3. Providing follow-up definitive testing and, if confirmed
4. Instituting appropriate management and services

It's phenotype based

• Deafness
• Cystic fibrosis
• Hypothyroidism
• Galactosemia
• Hemoglobinopathies
• Fatty(organic, amino) acid disorder: >40 disorders by Tandem Mass spectrometry
• Congenital Adrenal Hyperplasia
• Biotinidase deficiency
• Severe combined immunodeficiency (T-cell lymphopenia)

Carrier Testing #

Goal is to identify asymptomatic carriers with no family history of disease

Ethnicity based

Week 3: Next-gen sequencing for solving diagnostic dilemmas #

Whole Genome Analysis #

A genome-wide search for disease-causing variants

• Karyotype: Chromosomes under the microscope
• Cytogenomic array for large deletions/duplications
• Whole genome sequencing
• Exome sequencing

Clinical interpretation of variants #

Typical individual differences from reference

• ~5-10M SNVs (varies by population)
  • 40-100k SNVs in coding exons
    • 10-12k synonymous (no amino acid change)
    • 8-11k non-synonymous, in 4-5K genes
• 200-500k indels (1-1000 bp, varies by population)
  • ~150 in-frame indels in exons
  • ~200-250 shift the reading frame of an exon
• 500-1000 CNVs (>1000 bp)

Annotation of variants

• Basic
  • Gene name (if in a gene)
  • Chromosome location of the change (position in reference genome)
  • Location of the change within the mRNA/cDNA
  • Location of the amion acid change in the protein
  • Effect on protein
• Variant dependent methods
  • Allele frequency in general population
• Disease-dependent methods
  • Cosegregation of variant with disease in families
  • Prior association of gene with disease (OMIM, ClinVar, Locus Specific Mutation Databases)
  • Gene in a disease pathway

Predicting the effect of a variant is CHALLENGING

• Probably damaging
  • Stop-loss
  • Stop-gained
  • Frameshift
  • Splice disruptor
• Possibly damaging
  • Non synonymous
  • In-frame In/Del
• Likely not damaging
  • 5'/3' UTR
  • Synonymous
  • Intergenic
  • Intronic
  • Non-coding genes

Predicting the effect of non-synonymous variants

• Evolutionary conservation
• Protein structure
• Amino acid properties

Typical classification scheme

• Known pathogenic
• Likely pathongenic
• Variant of unknown significance (VOUS or VUS)
• Likely benign
• Benign

Using NGS for diagnostic dilemmas #

Dignostic Odyssey - delay and expensive

Intractable inflammatory bowel disease (WES)

• ~16,000 total variants, ~1,500 were "novel"
• ~7,100 non-synonymous substitutions, premature stops, ~1,100 were novel
• 136 variants fit autosomal recessive(AR) or X-linked(XL), 1 variant altered a conserved amino acid.

In a trio, if child affected,

1. Dominant: One copy of gene affected
2. Ressive: Both copies of gene affected
   • Homozygote
   • Compound heterozygote
   • X-linked
3. New mutation

Practical Aspects #

Incidentalome

• Unanticipated Pathogenic Variants

Week 4: Methods for Dissecting the Genetic Basis of Complex Diseases #

Background #

Complex diseases

• Mendelian: clear inheritance pattern, but complex no clear
• Phenocopies: Non geneic form of disease
• Incomplete penetrance: Not all genetically susceptible people develop disease
• Variable expressivity: Same genetic factor causes multiple phenotypes
• Genetic heterogeneity: Mutations in different genes can lead to same disease

Symptoms suggestive of a genetic condition

• Earlier age at onset of disease than expected
• Condition in the less often affected sex
• Family history with multiple generations affected
• Disease in the absense of known risk factors

GWAS methods #

GWAS approach

1. Genotyping: haplotype tagging
2. Study population: Cohort, Case control, Confounding and bias
3. Analysis and interpretation: Significance testing, Measures of effect, External validity

GWAS platform, SNP array

• 1,000,000 SNPs in one experiment
• Direct and indirect capture of 'all' common variants by using 'tag' SNPs
• Coverage of SNP array
  • Asians/Europeans: 67-89%
  • African: p46-66%

Study designs

• Common observational studies
  • Cohort
  • Case-control
• Common biases
  • Confounding
  • Misclassification bias

Race is a common confounder in GWAS

GWAS analysis #

Armitage trend test

Hypothesis testing and p values.

• What does a p<0.05 mean?: <5% probability that the observation is due to chance (i.e. a false positive) --> "statistically significant"

Need for correction for multiple testing

Manhattan plot showing genome-association

Resons for association

• True association (true positive)
• False association (false positive)

Properties of a valid association

• Not due to chance
• Free of bias
• Reproducible

GWAS interpretation #

Calculation of risk

• Risk = incidence of disease
• Can be calculated from cohort studies

Calculation of a relative risk

• Relative risk = ratio of two risks
• Measures the 'effect' of the variant on risk of disease
• xx fold increased risk of disease

In case-control study, we can calculate the ODDS of disease

• Odds ration = ratio of two odds
• xx fold increased odds of disease

Odds ratio vs. relative risk

• don't do a lot of epidemiological -> same thing
• rare disease -> approximate each other
• common disease -> Odds ratios overestimate relative risk

What do we know about the genetics of common, complex diseases? #

Missing heritability, limitaions of GWAS... what we're missing

• Common SNPs not tagged well
• Rare variants
• Other types of variants (CNVs, etc)
• Epistatic effects (Epistasis, gene-gene interaction)
• Effets of gene x environment interaction

Week 5: Clinical applications of genomics - Predictive Testing for Common, Complex Diseases #

Background, Analytic and Clinical Validity #

Predictive (predisposition) testing: predict whether a person likely to get disease

Genetic tests for complex diseases

• Familial
  • Breast cancer and Ovarian cancer (HBOC) - BRCA1, BRCA2
  • Colorectal cancer (Lynch syndrome) - MMR genes
  • Hypercholesterolemia (FH) - LDLR/APOB
• Thrombophilia - F5
• Hemochromatosis - HFE
• Celiac disease - HLA

ACCE Model: How the variance is ready for clinical use ( http://www.cdc.gov/genomics/gtesting/ACCE/acce_proj.htm )

Analytic validity: How accurately and reliably the laboratory assay measures the genotype (sensitivity and specificity), CLIA

Clinical validity: How consistently and accurately the test detects or predicts the disease

1. Clinical sensitivity: Among people with disease, how many test positive
2. Clinical specificity: Among people without disease, how many test negative
3. Positive predictive value (PPV): Among those with a positive test, how many will develop disease (penetrance) - BRCA1 mutation PPV is ~60%
4. Negative predictive value (NPV): Among those with a negative test, how many will remain disease-free - NLA-DQ test has 100% NPV for Celiac disease

Ciinical utility

• Does it improve health outcomes?
• Balance of benefits and risks

Case Scenario - Coronary Artery Disease Example #

Coronary artery disease (관상동맥질환)

Key concepts

• A genetic test is not necessarily measuring the causal variant, but rather, a genetic marker that may be linked to the causal variant.
• Results of genetic test may have utility in motivating changes in risk factors or improve medication adherence.
• In some cases, family history remains a better predictor of disease than any specific genetic test.
• In the U.S. most genetic tests are considered laboratory developed tests (LDTs) and are currently not regulated by the FDA.

Case Scenario - Diabetes Example #

Diabetes mellitus

Key concepts

• Multi-marker genetic risk panels should be interpreted with causion
• Genes identified to date poorly explain genetic underpinnings of disease
• Several resources exist to find objective information/guidance on tests
• Genetic testing should not distract from modifiable risk factors (diet, lifestyle, medication)

Case Scenario - Alzheimer's Example #

Alzheimer's disease

• GINA - what it does and doesn't cover
• Personal utility is increasingly being recognized as rationale for genetic testing
• Where to find information about available tests
• Utility of targeted variant analysis versus sequencing entire coding region

Week 6: Clinical applications of genomics - Pharmacogenomics #

Background - Genetic factors affecting pharmacokinetics and pharmacodynamics #

Adverse Drug Reactions (ADR): unintended and noxious, although individually rare, are collectively common

Pharmacogenomics: using a patient's genomics information to improve the efficacy and/or reduce the side effects of drugs

Pharmacokinetics: how the drug concentration changes as it moves through the body (ADME)

1. Absorption: orally, intravenously, inhaled
2. Distribution: circulatory system, transporters
3. Metabolism: enzymes
4. Elemination: urine, feces, breath

Many drugs are metabolized by the polymorphic Cytochrome P450 enzymes.

• UM: Ultrarapid metabolizer - extra copies
• EM: Extensive metabolizer (normal)
• IM: Intermediate metabolizer - one non-functioning allele
• PM: Poor metabolizer - two non-functioning alleles

Codeine metabolism

• codein ->(CYP2D6)-> morphine -> target
• PM can not convert codein to morphine, so higher dose required

Warfarin metabolism

• Active compound -> effect to target ->(CYP2C9)-> inert metabolite -> elimination
• PM can not convert to inert, so lower dose required (due to it's toxic side effects)

Pharmacodynamics: how the drug exerts its effect on the body (potency)

VKORC1, target of courmatin derivatives (e.g. Warfarin): varnant upstream leads to reduced expression

Off target effects - Avacavir hypersensitivity

• Drug affects target, but also interacts with unintended target
• Abacavir binds to host HLA-B in patients with HLA-B*5701 genotype
• Patients shoud avoid this drug

What pharmacogenetic tests are available? #

In FDA

• 158 drug-biomarker pairs
• 12% of 385 drugs approved 1998-2012
• Not all PGx markers in drug label are clinically valid
• Tumor markers is 34%
• CYP450s is 34% (CYP2C9, CYP2C19, CYP2D6)
• HLA genes increasingly associated with severe adverse events

PharmGKB - PGx biomarker levels

• Testing required
  • HLA - Carbamazapine
  • CFTR - lvacaftor
  • CYP2D6 - Tetrabenazine
  • OTC, POLG - Valproic acid
  • CYP2D6 - Pimozide
• Testing recommended
  • HLA - Abacavir
  • TPMT - Azathioprine
  • CYP2C19 - Clopidogrel
  • CYP2D6 - Dextromethorphan/quinidine
• Actionable
• Informational

Is my patient a candidate for pharmacogenomics testing? #

CFTR genotype-devendent efficacy of Ivacaftor

• Cystic fibrosis 85% is ▵508F
• W1282X
• G551D (4%) <- Ivacaftor effective only

Where to find?

1. PLoS Currents: Clinical validity and utility
2. GAPP Knowledge Base

Evaluating PPV and NPV of test

• Rare disease or outcomes can never lead to high PPV, no matter how good the sensitivity/specificity of the test

Where to get testing done and how to interpret the result? #

PharmGKB: Manually curated pharmacogenomics knowledge base including information from drug label, clinical testing labs and dosing guidlines.

GTR (Genetic Testing Registry)

CPIC (Clinical Pharmacogenomics Implementation Consortium)

Week 7: Clinical application of genomics - cancer management #

Background and cancer biology #

Cell division (mitosis) is regulated by cell-cycle controllers, growth factors and their receptors that induce or inhibit proliferation --> Cancer cells have acquired ability to bypass growth signalling and lose growth control.

Mutated genes

1. Oncogenes
2. DNA repair enzymes
3. Tumor suppressors
4. Chromatin modifiers

Proto-oncogenes

• Normally involved in cell cycle and growth signalling
• 'On' switch
• When mutated become Oncogenes
• Gain of function
• Promote proliferation

DNA repair during replication and exposure to carcinogens

• Excision repair
• Recombinational repair
• Mismatch repair

Tumor suppressors

1. Slow cell growth
2. 'Off' switch
3. Signal apoptosis of damaged cells
4. Mutated - loss of function

Chromatin modifiers - Epigenetics

• Mutations in epigenetic machinery
• Global hypomethylation
• Promoter specific hypermethylation

Hereditary cancer #

Familial Recurence Risk

• High risk: Testicular cancer, Thyroid cancer (8~9 times)
• Others 1~2 times

Features of hereditary cancers

• In the individual patient
  • Multiple primary tumors in the same organ or different organs
  • Bilateral primary tumors in paired organs or multifocality within a single
  • Yonger-than-usual age at tumor diagnosis
  • Tumors with rare histology
  • Tumors occurring in the sex not usually affected
• In the patient's family
  • First-degree relatives with same tumor history

Genetic basis of hereditary breast cancer

• First degree family history increases risk 2-fold
• Only 5-10% of breast cancer is hereditary
• Inheritance is complex at the disease level, but some individual genes behave in Medelian fashion
• BRCA1/BRCA2 - Hereditary Breast and Ovarian Cancer

Genetic basis of hereditary colon cancer

• Family history increases risk 2-4 fold
• FAP (Familial adenomatous polyposis): Mutations in APC gene
• HNPCC - Lynch Syndrome

Knudson's 2 Hit hypothesis for tumor suppressor genes

• Tumor suppressors are recessive at cellular level
• Inherited cancers behave as dominant trait

Should my patient undergo genetic testing?

Genetic testing for known cancer susceptibility genes

• A negative test result is no guarantee that cancer WILL NOT develop
• A positive test result is no guarantee that cance WILL develop
• Risk management strategies for a postive test
  • Surveillance
  • Hormone therapy
  • Lifestyle changes
  • Propylatic surgery

Tumor genetic landscape #

Measuring tumor gene expression

• One or a few genes at a time: e.g. HER2, ER, PR
• Genome-wide measures (Expression microarrays, RNA-seq)

Measuring somatic mutations in tumors

• Sequencing DNA from tumor-normal pairs

COSMIC database

How many genes are mutated in the average solid tumor?

• 33-66 genes in an average solid tumor
• Mostly SNVs
• Variable by cancer type

Drivers and passengers

• Driver mutations provide a selective growth advantage
• Passenger mutations have no effect on neoplastic growth
• How to identify driver genes?
  • Mountains and hills (frequency)
  • Patterns

How many driver genes exist?

• 3284 tumors sequenced
• 294,881 mutations reported
• 125 mutation driver genegens identifid
  • 71 tumor suppressor genes
  • 54 oncodones

Metastatic disease

Clinical applications - prognosis and treatment response #

Breast cancer prognostic markers

• Oncotype DX
  • A 21-gene expression score (16 prognostic genes and 5 housekeeping)
  • Scale of 0-100, strata of low, intermediate or high risk
  • Predicts 10-year risk of distant recurrence in Estrogen Receptor (ER) positive breast cancers (may benefit from adding chemo to their hormone treatment)
  • Predicts responsiveness to CMF (Cyclophosphamide, Methotrexate and 5-Fluorouracil) chemotherapy
• MammaPrint
  • A 70-gene expresison profile
  • Regardless of estrogen receptor (ER) status, with tumors of less than 5 cm
  • Distinguishes those predicted to have good prognosis (no relapse within 5 years) from poor prognosis (relapse within 5 years)

HER2-positive breast cancer

• The HER2 gene is amplified in 20% of breast cancers
• Referred to as HER2-positive cancers
• Make more HER2 protein than HER2-negative cancers
• The extra HER2 protein causes increased signal pathway activation, which contributes to the uncontrolled growth and survival of these cancers.

Chronic myelogenous leukemia (CML)

• Blood cancer caused by reciprocal translocation (Philadelphia Chromosome), resulting in oncogenic BCR-ABL gene fusion
• BCR-ABL found in 95% of CML
• Multiple targeted medications have been created which specifically inhibit this oncogene (imatinib, dasatinib, nilotinib)
• Previously median survial 4 years, now 20-25 years

KRAS mutations in Colorectal cancer

• Anti-EGFR therapies cetuximab (Eritux) and panitumumab (Vectibix) don't work in KRAS-mutated cancers

Vemurafenib and malignant melanoma

• Mutant BRAF (~50% of melanomas)
• BRAF inhibitors (Vemurafenib) block constitutive activation of downstream pathway to decrease cellular proliferation

Different cancers may share common genetic alterations

• 60% of melanomas respond to the BRAF inhibitor vemurafenib
• 100% of hairy cell (HC) leukemias response to BRAF inhibitor
• 5-10% colorectal cancers no response to BRAF inhibitor

Challenges in Molecular Profiling fo rTargeted Treatment

• Bioinformatics
• What does it mean when a mutation normally associated with an inherited cancer is found in a tumor sample?
• Tumor heterogeneity
• Tumor targets identified but therapy not approved or reimbursed for that indication

Advances in cancer genomics

• Continued development of targeted treatments
• New treatment paradigms - immunotheraphy
• Improved diagnosis - liquid biopsies

Hyungyong's retrospective #

제출 시간을 잘못 알고 있어서 퀴즈 제출 하나 놓치고는 이수 못하는 거 아닌가 걱정이 많았다. (PM 12:00은 밤이 아니라 낮임) 다행히도 간신히 통과함. Certificate 임상에서의 Persnolized medicine에 대한 체계적인 소개를 받을 수 있었다는데 큰 의의가 있었다. 특히 암유전체에 대해 몰랐던 것들이 많았음. Coursera 시스템도 꽤 괜찮았는데, 특히 이수 후, 클릭 한번으로 Linkedin 프로필에 바로 등록되는 건 매우 이채로왔다. 기분도 매우 좋음. 여세를 모아 꾸준히 실력향상~ --Hyungyong Kim, 2014-06-06

Incoming Links #

Related Articles (Article 0) #

• Exome sequencing
• Human genome

Suggested Pages #

• 0.260 개인유전체 검사 결과를 설명하는 방법
• 0.156 Cell-free tumor DNA
• 0.087 잃어버린 유전 가능성(Missing heritability)에 대해
• 0.050 Microbiome
• 0.045 CODEX: a normalization and copy number variation detection method for whole exome sequencing
• 0.024 2010
• 0.024 WGS
• 0.022 2016
• 0.021 Whole-exome sequencing and clinical interpretation of formalin-fixed, paraffin-embedded tumor samples to guide precision cancer medicine
• 0.020 CODEX
• More suggestions...