Precision medicine Flashcards
(91 cards)
1
Q
what is precision medicine?
A
medical care designed to optimize efficiency or therapeutic benefit for particular groups of patients, especially by using genetic or molecular profiling.
within the last 10-20 years we have moved towards targeting the therapy to those that will get the most benefits – by examining their characteristics in detail, giving them a personalised treatment where the benefits are most significant and side effects are minimised
2
Q
what is the traditional approach to treatment?
A
The traditional approach is treating the population with the same treatment – a mixed group but we target the average patient – measure if it is effective by looking at our outcomes, some would develop no outcome, other negative but normally on average most would have improved outcome
3
Q
whatshe difference between precision medicine and personalised medicine?
A
- Precision matched to a population characteristics (genome, environment, population health data then data analytics)
- Not interchangeable – personalised is unique to the individual (and is unlikely to work for anyone else)
Enabled by new omics technologies, Mostly genetic markers currently, Particularly now pharmacogenetics (the study of how a persons genetic makeup (genome) influences their response to treatment)
4
Q
how many drugs currently use pharmacogenomic markers?
A
Currently > 200 drugs have label information regarding pharmacogenomic markers* Incl. CVD, lung disease, HIV, cancer, arthritis, high cholesterol and depression
5
Q
how can we target mutated proteins?
A
Next gen sequencing allows identification of single gene mutations which cause disease. Gene therapies can rectify these mutations in some cases.
If genetic mutations can be rectified, it is possible to target mutated proteins with small molecule inhibitors or Mabs
6
Q
give some examples of precision medicine used for cancer currently
A
- Monoclonal antibodies like Herceptin is a form of precision medicine - HER2 mutation Different responses and outcomes to Cx and radiotherapy
- Drugs targeting BCR-ABL via tyrosine kinase (TK): Small molecule TKI)
- Anti-angiogenesis treatments, EGFR inhibitors, BRAF inhibitors (if + test for BRAF mutations)
- EGFR inhibitors (test for mutations), ALK inhibitors, ROS1, KRAS, NTRK, BRAF, MET, RET, anti-angiogenesis
- Mabs against CD20, CD19, CD79b, often conjugated to chemotherapy drugs
- BRAF, MEK, KIT inhibitors after identifying mutations
7
Q
precision medicine may help by/when…
(3)
A
- Stratifying diseases to identify improved std therapeutic outcomes
- Targeting via known mechanisms
- Targeting by location
8
Q
what factors complicate the question are tumours self or non self?
A
- Some harbour microbes (viruses/bacteria/fungi)
- Some are caused by viruses
- Some lose HLA
- Expression of Tumour Associated Antigens (abnormal quantity, location or time) – e.g. NY-ESO-1, XAGE1, MAGE, CTA
- Expression of Tumour Specific Antigens (resulting from DNA mutations)
9
Q
what is the spontaneous response to tumours?
A
10
Q
which MHC class are tumour antigens expressed on
?
A
- Tumour proteins are endogenous and could be presented by tumour cells on HLA-I
- Dead tumour cells can be phagocytosed by APCs (Mf, DCs) as exogenous proteins on HLA-II, and presented on HLA-I (cross-presentation)
11
Q
what is the principle of immunotherapy?
A
aim to enhance the natural anti-tumour response
12
Q
give five examples of immunotherapies
A
- Immune checkpoint inhibitors (PD1/PDL1)
- T cell transfer therapy (CAR-T cells)
- Monoclonal antibodies (anti-TAA mAbs)
- Treatment vaccines (TAAs)
- Immune system modulators (cytokines, BCG/oncolytic viruses, immunomodulatory drugs e.g. angiogenesis inhibitors)
13
Q
how can we boost the anti-tumour T cell response with immunotherapy?
A
immune checkpoin inhibitors like PDL1/PD1
they prevent t cell exhaustion, removing the checkpoint to allow t cells t o remain active against the tumour
14
Q
what is the sceintific basis of immuhne checkpoint inhibitors?
A
- Acute stimulation of T cells requires rapid response, closely followed by immunomodulation (PD-1/PD-L1) to prevent uncontrolled T cell activity
- Chronic stimulation of T cells leads to “exhaustion” (PD-1 Tim-3, LAG on T cells)
- Tumours have an immunosuppressive/exhausted environment, incl. PD-L1 on tumour cells
- Immunotherapy interferes with PD-1/PD-L1 interaction, removing checkpoint
15
Q
what are the types of mutations?
A
▪ Substitution (point mutations) :
Silent (same amino acid), nonsense (stop codon), missense (diff amino acid)
▪ Insertions
▪ Deletions
▪ Duplication
▪ Inversions
▪ Translocations
16
Q
give an example of a mutation leading to loss of function
A
p53 is a driver gene
* Mutated p53 leads to LOF
* As function is repressing cell proliferation in response to stress, LOF predisposes to cancer
17
Q
what causes tumours?
A
- DNA changes in Proto-oncogenes, Tumour suppressor genes, DNA repair genes lead to tumour growth
All driver genes - DNA changes induced by:
➢Environmental damage (UV, smoke, alcohol)
➢Inheritance
➢Viruses
➢Cell division errors
18
Q
how are mutated proteins recognised?
A
T cells identify peptide from non self proteins but they have to be presented on host MHC
During infections, non-self is obvious – viral/bacterial proteins are inside the cell
On tumours, non-self could mean proteins not normally produced in adult tissues – CTAs, splice variants etc.
Or abnormal proteins arising from cancer-causing mutations
19
Q
why dont we know what the t cell targets?
A
TIL analysis suggests 5-20 clonotypes in a tumour
* Neoantigens are peptides derived from mutated proteins capable of binding to MHC
* Cannot easily identify T cell targets as the TCR does not inform the MHC peptide
20
Q
breifly describe whole genome sequencing using Illumina
A
DNA fragmented then
▪ Adaptor sequences ligated to ends (contain primer sequences)
▪ Size sorted (size depends on machine)
‘Bridge amplification’
▪ Single strand binds to sequences attached to a solid surface
▪ Free end binds to a nearby complimentary sequence (bridge)
▪ dNTPs (unmodified) are added to create a double strand
▪ Denatured to form 2 strands attached to the tile
▪ Repeated to form local clusters of copies of the same sequence
21
Q
describe how the sequence is determined in illumina sequencing
A
▪ “Reversible terminators” instead of a mixture of dNTPs and ddNTPs
▪ Mutant DNA polymerase required to incorporate modified bases
▪ 3’ end free to incorporate next base
▪ All 4 modified dNTPs added
▪ Correct base incorporates
▪ Multiple bases cannot be incorporated due to blocking group
▪ Free bases washed away
▪ Laser excites fluorescent dye
▪ Base (colour of fluorescence) is detected
▪ Fluorescent dye and 3’ blocking group cleaved
▪ Cycle repeats
22
Q
why does the quality decrease with as the sequence is read in illumina sequencing?
A
The dna pol that adds these adapted nucleotides (mutant dna pol needed bcos they are flourescent) it doesn’t have a high fidelity – so the quality at the start is quite high but less and less reliable as you go along the fragment
▪ Initially just ~35bp of sequence now up to 300bp
▪ Modified DNA polymerase required to attach reversible terminators, can lead to errors in incorporation
▪ Sequence quality reduces towards the end of the read, as clusters get out of sync
23
Q
what is the most common implementation of illumina sequencing?
A
Paired–end Illumina sequencing
23
Q
what is the most common implementation of illumina sequencing?
A
Paired–end Illumina sequencing
24
describe the principle of paired end sequencing?
▪ Both ends of a sequence fragment are sequenced in turn
▪ Pairs are identified by shared location on the chip, producing 2 sequence files eg. sample_1.fq and sample_2.fq that are in sync.
▪ Then aligned to the reference sequence
25
what is in a FASTQ file?
Illumina sequencing results
multiple paired end sequences are aligned over a reference sequence
the file indicates the quality, you have a string of bases that you belive to be the sequence and below each position is a character indicating the quality
▪ Four lines per sequence
1. Sequence identifier (starting with @)
2. DNA sequence
3. ’+’ optional sequence identifier
4. Corresponding quality scores (single ASCII character per base)
26
what is phred?
▪ Quality score (Phred) is an integer representing the probability that the corresponding base is correct
▪ Phred of 20 = 1 in 100 or 99% accuracy
27
what can be sequenced by Next Generation Seqencing for mutanome analysis?
▪ Whole genome sequence (WGS)
▪ Exome (Exome-seq)
▪ mRNA
▪ miRNA
▪ Methylation sequencing
▪ Chip-seq
▪ Ribo-seq
28
what is the exome?
An exome is the sequence of all the exons in a genome, reflecting the protein-coding portion of a genome (a subset of the genome, exons only). In humans, the exome is about 1.5% of the genome.
29
what percentage of mutations causing disease are located in the exome?
85%
estimated
30
why would you sequence the exome?
* Next Generation Sequencing makes sequencing the exome a cost effective strategy
* Can screen all genes at once
o ~20,000 genes (average of 8 exons per gene)
o >150,000 exons
o ~50Mb of sequence
* Much cheaper and easier to interpret than whole genome
o Focuses on the part of the genome we understand best, the exons of genes
* Applications:
o To discover variants involved in rare Mendelian and complex diseases
o To detect somatic mutations in cancer
o Molecular diagnostics
## Footnote
(costs £600 to sequence as opposed to £1000 for entire genome)
31
how do you do exome sequencing?
Still taking genomic dna and fragmenting it up
Hybridisation
Wasg
Incubating it up with beads that will binds to sequences that we already known and pull them down
Captured DNA is then sequenced and analysed
32
how is exome sequencing analysed?
Start with QC: (quality control)
Remove adaptor sequences (we know what they were so computational remove)
Remove low quality sequence information (look at the fastq file and remove all with low quality)
Align with reference genome and start to compare (computationally)
so you can say according to probability, whether the base inserted is correct for the sequence, whether it is hetero or homogenous
use database to compare
33
what are the three stages of exome sequencing analysis?
Alignment
- raw sequence reads
- human genome reference sequence
Variant calling
- 'pile up' reads
- identify variants
Annotation
- annotated output files
34
how many variants are found per sample?
~23000 variants per sample
35
what are the three parts to the annotation stage of exome sequence analysis?
1. Gene and variant information (genes already known)
Gene name, Transcript ID, Exon number, Nucleotide change, amino acid change, Type of mutation; indel, missense, nonsense, silent
2. Cross reference with databases of known variants
(100000 genome project to show potential variation)
3. Functional prediction (is it involved in disease, could it be?)
36
what are the types of mutations found when exomes/genomes are sequenced?
▪ Structural mutations (usually>50-bp) – inversion, translocation, deletion, duplication, insertion.
▪ Indel short (usually) insertion or deletion. Sometimes leads to frameshift. Usually drastic alteration of translation product and targeted decay of mRNA.
We are focussing on single nucleotide variants
▪ Non-sense (stop gain) – change of single nucleotide causes premature termination of protein sequence. Change or loss of protein function.
▪ Silent (synonymous) – change of single nucleotide does not change the amino acid sequence. No change to protein function. (due to degenerate quality of the DNA)
▪ Mis-sense (non-synonymous) – change of single nucleotide changes the amino acid sequence. Potential change or loss of protein function or change to structure.
37
why is the variant allele frequencing in tumours often less than you think it will be?
when you sample a tumour you might not have all tumour cells - there are other cells (eg immune, fibroblasts, endothelial) that are contributing the normal genome
so the percentage might not be as high as you think becuase you only exclusively sequencing thte tumour genome
## Footnote
eg variant allele frequency in oesophageal cancer found to be 15%
38
how do you get deep sequencing coverage?
you look at about 100 times depths - each nucleotide is looked at 100 times so we know we are getting relable info
the fragments layer over each other so many times
39
how do you get deep sequencing coverage?
you look at about 100 times depths - each nucleotide is looked at 100 times so we know we are getting relable info
the fragments layer over each other so many times
40
RNAseq _ and _ are generally lower than Exomeseq
RNAseq coverage and quality are generally lower than Exomeseq
41
what can RNAseq in neoantigen discovery pipelines (transcriptomics) be useful for?
Can be useful for confirming the presence of a mutation and transcription of its gene in a tissue
42
describe the process of RNAsequencing for neoantigen discovery
PolyA and RNA captured
RNA fragmented and 5' and 3' adapters ligated using degenerate guide adapter
guide sequences removed. Reverstranscription primer anneals and first strand of cDNA synthesised
cDNA is amplified by PCR and adapters
emulsion PCR performed to covalently bind fragment to microbead
clonal template cDNA ready for sequencing from A towards the mcirobead
43
what are the issues for transcriptomics?
Transcriptomes are dynamic so get overrepresentation of genes that are preferentially expressed
Depth in transcriptomics is never a s good – RNA is labile prone to breaking down, not conducive to deep sequencing studies
coverage and quality is generally low
44
we can use sequencing to identify tumour mutations in tissue like?
* Identify DNA mutations in germline (PBMC)
* Identify DNA mutations specific to tumour
* (Sometimes) derive or confirm mutations using RNAseq
45
where can germline DNA be found for sequencign to identify tumour mutations?
what is wrong with sourcing the dna from here?
Germline dna found in circulating blood cells – not actually germline just fairly close.
A lot of tumours happen due to environmental stimuli so the germline dna is not going to reflect all of the things/mutations that have happened in your life
Better is to look at adjacent healthy tissue, should have all of the acquired mutations but the tumour should have these plus the additional ones that have caused tumourigenesis
46
how do you find tissue specific mutations?
compare healthy tissue with germline
47
why should you source healthy tissue adjacent from the tumour site to identify tissue specific mutations?
because different tissues have different mutational loads
due to exposure with the environment (eg skin, lungs, oesophagous)
48
which tissue had the highest mutational burden? and why?
Skin has the highest mutaional burden because it is constantly exposed to UV rays from the sun
## Footnote
then oesophagous and lung because are exposed to environment (smoking pollution alcohol food)
49
at most how many mutations can be associated with one cancer
400
this is seen in melanoma
(mostly passenger mutations)
50
how many driver mutations are assocatiated with cancer?
1- 12
51
which type of cancer has the highest mutational load?
Melanoma has over400 average (more c to t in skin cancers than any other cancer types (bcos this is what is associated with UV light)
52
what specific mutation is associated with exposure to UV light?
c to t
53
what specific mutations is associated with smoking?
c to a transversion
54
whats the difference in function between HLA1 and HLA2
HLA-I processing is of intracellular/cytosolic proteins in the ER (Self + viruses/bacteria/abnormal) – All cells
HLA-II processing is of extracellular/membrane proteins in phagosomes or lysosomes (self + bacteria/yeasts etc) – Mostly APCs
55
what recent discovery has changed the way we think about t cell activation and the tumour immune response?
we need both HLA1 and HLA2 presentation pathways to be able to stimulate a good t cell response
(because they stimulate different t cell subsets)
56
HLA genes are polymorphic and polygenic
what does this mean?
57
why are there so many potential HLA combinations?
Natural selection, has chosen this number to maximise survival – people who don’t have enough are prone to cancer – it is effective
58
to be neoantigens, peptides from TSA must bind to _ to be...
HLA
to be recognised by CD8 T cells
59
what is the most common HLA
HLAa0201 present in 40% of people in the world
60
the core of HLA binds what number of amino acids
9
61
what is the difference in binding capacity of HLA1 and HLA2?
* Optimum HLA-I peptide length is 8-11 amino acids (mostcommonly 9)
* Optimum HLA-II peptide length is 12-18 amino acids (mostcommonly15)
62
the receptors on t cells that detect hlas are specific for binding what?
the complex of the peptide bound to HLA
(has to recognise both parts to become activated)
63
what is the optimum peptide length for HLA1?
8-11aa
64
what is the optimum peptide length for HLA2?
12-18aa
65
how do HLA1 and HLA2 work together?
* Some in-vivo evidence that CD4 assistance is necessary for effective anti-tumour activity.
* Tumour cells can be MHC-II negative but still require CD4 assistance to be killed.
* Priming increases CD8 T cell clone numbers and increases their killing capacity.
* So it is desirable to find mutated peptides (neoantigens) presented by both HLA-I and –II – if they work together
In most cases neoantigens presented on hla2 to cd4 t cells – this causes release of il2 which activates cd8 t cells
66
what are HLA binding prediction algorithms - how are they use?
* Using real-world data, machine learning algorithms (e.g. NetMHC) can predict whether peptide sequences of a certain length would bind to HLA-I or –II of any known allotype with good accuracy.
* Take all protein coding sequences containing a mutation, chop them up into rolling e.g. 9 and 15 amino acid lengths and then predict if they will bind to any HLA-I or –II allotype.
* Computationally intensive as there are often millions of potential peptides to predict in a tumour mutanome containing 400 mis-sense mutations
## Footnote
Use sequencing info then machine learning algorithms, look at mutated region and see which hla they are likely to bind (due to the amino acid probability binding graphs)
67
what is a major problem with using sequencing info and machine learning algorithsm to predct what peptide HLA will bind?
They predict a lot of peptides that are never actually seen by the HLA
Because they have to be produced, recognised, cleaved, taken up and presented and we dont know if this happens or not invivo
68
rather than using just sequencing data to predict neoantigen binding to HLA what is a better way to approach the prediction?
Combining HLA typing data, mutation analysis and gene expression with HLA binding algorithms allows us to try and predict neoantigens in any given tumour sample
* The pVACseq pipeline attempts to combine peptide HLA binding affinity with gene expression data to predict if a mutated protein can make a neoantigen
69
How good is prediction of HLA neoantigen binding?
About 1-5% of predictions lead to a T cell response.
70
why do the percentage of predictions so rarely lead to a t cell response?
✓ Easy to predict if gene is transcribed
✓ Harder to know if protein is actually produced
✓ Even harder to predict if it can be processed by HLA machinery
✓ Harder still to work out if the peptide is different enough from the wild type version to stimulate an immune response
✓ Remember: T cells recognising self peptides are selected out of the T cell population to prevent autoimmunity
71
what ways can we measure t cell response to our predicted neoantigens?
Tetramer – quantifying T cell numbers
Elispot
flow cytometry to access t cell activation
72
how can you use tetramers to quatify T cell numbers
looking for how many t cells in the blood stream will recognise the peptide
* Peptide mixed with MHC-I or –II molecules has a fluorescent tag and is incubated with T cells.
* Binds to the cognate receptor.
* pMHC complex binding to specific T cells by flow cytometry identifies neoantigen-specific T cells
* Outcome: % of T cells capable of binding that peptide
73
how can you use elispot to measure the t cell functional response?
deliver peptide to a dish full of blood (contains monocytes that will take up the peptide and t cells that it will be presented to)
if t cells are activated (/peptide stimulates t cells) they will secrete interferon gamma (IFN-g) which will be detected if found on the bottom of the plate)
you can have only cd4 or only cd8 or both in there to see which one they stimulate
* Very sensitive
* Use blood from same donor (matched HLA type)
* Either “restricted” peptide or long peptide
* Compare WT to mutant peptide
* Outcome: Number of IFN-g secreting T cells/million starting population
74
even though it is a straightforward method what are the downside to elispot?
Such a wide array of t cells in body that they are likely to recognise it – however in biological situation they may never even be processed
So far removed from the person – will they ever do this is real life bcos you are delivering peptide
75
how can flow cytometry be used to assess t cell activation?
take cell, give epitope then use stain to detect inerferon gamma and create a flow cytometry graph
* 4-1BB and Ox40 are T cell activation markers
* Expose T cells to neoantigens them measure activation
* Flow cytometry can be used to measure the cell surface expression of proteins such as these activation markers
* Outcome: % of specific T cell population expressing activation markers
76
what are tandem mini genes?
A tandem minigene construct comprised 6 to 24 minigenes that encoded polypeptides containing a mutated amino acid residue flanked on their N- and C-termini by 12 amino acids. Tandem minigene constructs were synthesized and used to transfect autologous APCs or cell lines co-expressing autologous HLA molecules.
77
how can tandem minigenes be used to asses neoantigen t cell activation?
78
how can tandem minigenes be used to asses neoantigen t cell activation?
79
for neoantigen therapeutics, what vaccine design options are under development?
DC-based vaccines
SLP vaccine
RNA vaccine
80
what are DC based vaccines?
81
what are SLP vaccines?
82
what are RNA vaccines?
83
how are we currently using predicted neoantigens for vaccines?
what is one problem with this?
Administer with pd1 therapy, to make sure the t cells are active to be able to recognise this peptide that you are vaccinating with
This confuses whether it is really effective or not
84
vaccinating with predicted neoantigens (and checkpoint inhibitors) saw what positive effect in melanoma?
At the moment only works with a few people
Melanoma saw complete remission with 10-20%
85
why are vaccines from predicted neoantigens not commercially available?
the method is laborious
effectiveness is not cconfirmed yet (as currently administered with pd1 therapy)
86
what are the potential problems with predicted neoantigens vaccination?
* Tumours have an immunosuppressive environment
* They are heterogeneous, with different regions containing different mutations ( they don’t all contain the neoantigen, if all are killed then it could lead to more aggressive cells to grow back because have elimintate the immunostimulating peptides cells)
* T cell killing of cells expressing a specific neoantigen leads to immunoediting
* Only ~10% of mutations are in driver genes, so most are unique to the individual
* some solid tumours are such that T cells can’t infiltrate (desert)
* Often requires checkpoint inhibitor therapy to work
* Best vaccine formulation currently being investigated in trials
87
how can car t cells therapy be create from predicted neoantigens
If the T cell receptors responding to neoantigens are cloned, CAR-T type therapies could also be considered
88
what is a new way of predicting/determining neoantigens?
New studies attempt to directly observe neoantigens being presented by the tumours using mass spectrometry
89
how are HLA polymorphic and polygenic?
what is the difference?
Polygenic: Made up of multiple genes. The HLA is polygenic since there are more than 20 different class I and class II genes in the complex.
Polymorphism: Presence of different alleles of a gene at a frequency of more than 1%
90
the average person has how many variants in their genome
23000
