Class 12 DNA sequencing and cancer

DNA pol error rate ~10-9 per base copied

How many errors in a “typical” somatic cell?

Most errors don’t have detectable effects

But some errors do: oncogenesN – dominant if “activated”tumor suppressor genesN – recessive

2-hit hypothesis in inherited cancer syndromes, BRCA, FAP

loss of heterozygosity in tumor DNACancer nowadays viewed in molecular-genetic terms

Implications for therapy

Can inhibit some overactive oncogenes with small molecule inhibitors (imatinib, etc) often act intracellularly or with antibodies to cell surface receptors

(panitumimab, etc) that act in pathwaysthat stimulate intracellular oncogenes

But can’t replace function of inactive suppressors

Example of pathway activating oncogenesExtracellular ligand(epidermal growth factor,EGF) binds EGF receptor,which binds another protein, which causes cytoplasmic tail of EGFR to get phosphorylated,which activates otherproteins (here including Ras oncogene)…which turn on other genesthat stimulate cell growth. Antibody to EGFR may stop process, but if Ras is mutated and constitutively active, Ab to EGFR won’t work because Ras is “downstream”Image from Google search “egfr kras signaling pathway”

Kras mutated and constitutively active in ~40% ofcolon cancers

Large effort has gone into whole genomesequencing of tumors and comparisonto non-tumor DNA from same patient

What are main results?several hundred oncogenesseveral hundred tumor suppressor genesorganized in at least tens of pathways

Tumors are “clonal” but continue to acquire mutations

When you sequence a tumor, do you get sequence ofmajority of cells or of individual cells, withunique mutations?

What are “driver” vs. “passenger” mutations?

What are some clues to identifying driver mutations?

occurrence in multiple tumorsmutated in inherited cancer syndromes

Do you think there a more tumor suppressor mutationsor oncogene mutations driving tumors? Why?

How fast do tumors grow?cell birth rate b (# divisions/day, ~1/few days)

balanced by cell death rate dcell doubling rate k, N(t)=N02kt

k related to b-d

More ways to inactivate a gene (stop codon nearly anywhere) than to make it overactive, so suppressor mutations should exceedactivating oncogene mutations, but need toinactivate both copies of a suppressor, soanswer not obvious

Types of cancer therapysurgery – curative intent or for palliationradiationchemo to kill rapidly dividing cells

-> toxicity from killing normal rapidly dividingcells in gut, bone marrow, skin

drugs or antibodies that target oncogenescould be more specificbut still often have major side-effectsexamples – antibody to EGFR (drug names

ending in “ab” are antibodies) small drug inhibitors (drug names

ending in “ib” are inhibitors)

Problem of “development” of resistance to chemo

Roles of DNA sequencing

Research – find what genes are involved in cancerbig challenge – interpreting changespassenger vs driver mutationsare mutations in non-coding regions (98.5% of total)

important?which mutations in coding regions are relevant?

Patient carewhich genes are mutated in a specific tumor?is whole genome seq. necessary or would seq. of ~hundred known oncogenes and suppressors do?

Patient care – cont’d.

diagnostics – circulating tumor DNA akin topre-natal dx from circ. fetal DNA

? useful for screening or just dx of already ill? use to follow treatment – ? more sensitive

than other biomarkers, e.g. CEA, PSAdo genetic assays need to be specific for individual

patient’s mutations or are mutationssufficiently common that “generic” tests ok?

“Beaming” assay emulsion pcr for particular oncogenes

-> copies single templates on beadsbreak emulsion, hybridize flourescent oligo probes

to beads, different colors for oligosmatching wt, mutant, and common seq.

determine bead color with flow cytometry

http://openwetware.org/wiki/Image:Flow_cytometry

pre-op day 3

day 48 day 244

What is plotted?What do #sin quadrantsindicate?

Beaming assay for Kras mutations from Vogelstein

How sensitive is assay to mutations occurringin fraction of tumor cells as tumors evolve?

What fraction of circulating DNA is from tumor? How many beads can you assay?

Use of sequence info in therapy

possibly to identify unexpected mutations (e.g. uncommon in patient’s tumor type) that might suggest use of different drug – this is hypothetical

identify drugs unlikely to be effective – e.g. Ab to EGFRin pts with oncogenic Kras mutations

Use of sequence info in therapy

relevance to patients – avoid (often severe) toxicityin patients in whom drug won’t work(panitumimab has lots of toxic skin, gut effects)

relevance to payors - @$1000’s/dose, cheaper to genetest everyone to avoid use when predictably ineffective = “companion diagnostics”

relevance to pharmaceutical companies – use in resistant patients weakens evidence for efficacy, lack of efficacy is major cause of failure to get FDA approval

Questions from this paper

How fast do tumors (cells resistant to chemo) grow?How sensitive are tests for tumor mutations?What is normal mutation rate?What is probability that particular oncogene mutation

has occurred?How many mutations -> drug resistance?Do resistance mutations pre-exist in tumors, explaining

usual drug failure after few months?Implications for multi-drug therapy

How would you describe the patients in this study?

What is progression-free vs.overall survival?

Does prior Kras mutation predict poor response?How long before progression in those w/acquired

Kras mutations?

What do panels show?

Do mutationsor CEA ortumor sizeassays predict treatment failure sooner?

What is doubling rate?

patient 1 patient 2

If doubling time t is ~10d and progression time T is ~150 weeks

how much has mutant cell # increased in time T?

N/N0 = 2T/t = 215 = 3*104

How much circulating DNA?How many cell equivalents in 1ml @6pg/cell? What fraction f is from tumor cells vs. normal cells?

What are these plots?

wtmutant

How many dots?What is the lowest % (or number) mutant detectable?Suppose 1 mutant dot is reliable and 105 dots ->min fraction of mut. tumor cells detectable = 1/(f*105)If f = 0.1%, 1% tumor cells is min detectable

How many tumor cells in a 100mm2 (x-ray) tumor

Tumor vol = (area)3/2 = 1000mm3

Cell vol ~ (10mm)3 => 109 tumor cells

If 1% are mutant when mutation first detected, howmany were there before panatumumab was started?

107/(3*104) = 3*102

Is this consistent with expectation if DNA polmakes 1 base error every generation andyou have 109 cells => 109 genomes copied?

-> ~1 error in every position

If 42 positions confer resistance to panitumumab(their estimate), expect ~40 mutant cells to pre-exist; not too far off estimate of 300 givenlarge variance in rates of doubling, etc.

If 40 (or 300) mutant cells are expected to bepresent, on average, by chance in small tumor,what is probability that a tumor has no such cells?

Poisson distribution pi=e-mmi/i!pi = probability of a tumor having i

when average number/tumor = mp0 = e-m = e-40 or e-300 = 10-18 or 10-131

What is chance that at least 1 cell in tumor with 109 cellshas oncogene mutations conferring resistance to 2 different drugs, if the mutations do not overlap and changes at 40 positions confer resistance to each drug?

(40/109) * (40/109) * 109cells @ 10-6

Implication – multidrug therapy might avoidoutgrowth of resistant mutants

Main ideas

Mutations in cancer cells drive growthgain of function = oncogenesloss of function = tumor suppressor genes

Some drugs target oncogenes by binding to themor their partners in cell signaling cascades

Mutations conferring resistance to individual drugs likely preexist in tumors becausethey contain large numbers of cellsharboring mutations just on basisof DNA pol error rate

Multidrug therapy targeting different oncogenes/pathways might overcome these resistancemechanisms, but …

DNA sequencing has been important for discoveryof different mutations driving cancer

Often difficult to determine if individual mutationsare drivers or passengers

Genotyping specific genes in patient tumor DNAs tosee if most tumor cells already carry resistance-causing mutations can prevent futile use of expensive toxic drugs

Not clear if routine sequencing of exons orwhole tumor genomes is useful clinicallyat present, as opposed to targetedgenotyping or sequencing

“Beaming” is nice use of emulsion pcr and flowcytometry to detect not too rare mutations intumor cells

HAPPY THANKSGIVING – work on picking a topic forstudent presentations beginning 11/30