Direct to consumer Genetic testing week 9 Flashcards
why do we care about direct to consumer testing?
No regulation of the industry, genotyping can be incorrect, up and coming industry with huge explosion of companies offering the tests
Regulation of DCT in the UK and America
UK - no regulation
American - FDA starting to step in and regulate the industry
Case report showing inaccuracy of information reported by DCT’s?
Case report showing a number of people who had DCT and been told they were immune to norovirus and subsequently turned out that everyone in the family then tested positive for norovirus except for 1 person.
Case report showing inaccuracy of information reported by DCT’s - what does this show?
this shows that either the genotyping was incorrect or that the genotype was not responsible for resistance to norovirus so need to bear in mind that though someone has been told something, it doesn’t mean it’s true
What is Direct to consumer testing?
Companies use data from GWAS. CaseControl studies examining SNPS
What does GWAS do?
GWAS compares large groups of individuals (unaffected controls vs individuals with symptoms of a specific disease) in an attempt to distinguish between non-harmful changes in the DNA code and pathogenic, disease- causing/predisposing changes
GWAS revision
Case-control study of common genetic variant. Patients and controls, take DNA run them on SNP chips, compare differences and find SNPS associated with disease by statistical testing
How can these companies use GWAS to make a test?
Following GWAS the companies will have a list of associated SNPS and conditions, you can put it into a report, test people and report back to people what it is they’re more susceptible to developing.
BAsic process of DCT
Buy a kit, Spit into tube, Send away for analysis, Receive results a few weeks later
SNP chips - market leader?
Illumina SNP chip - market leader
How do SNP chips work?
- Visualising what’s seen on the chip
- Hundreds of thousands of different SNP probes and the varying colours depending on the genotype of that individual person
- High resolution camera will take a picture, analyse the pictures and make genotyping calls based on the colours of spots
Additive risk prediction
SNP based susceptibility is usually predicated on an additive risk prediction so the more risk alleles you have the higher your risk is
Additive risk prediction: Odds ratio numbers
o zero alleles = slightly protected,
o up to two risk alleles = you’re at the normal risk level,
o six or greater risk alleles = up to maybe seven or 8 odds ratio
ADditive risk prediction odds ratio numbers in context
Population wide things, when we think about traditional risks i.e. lung cancer. o If you’re a smoker over 30 cigarettes a day your odds ratio for lung cancer would be somewhere around 100
Overwhelmingly in some of these SNPS association
looking at smaller odds ratios - looking at susceptibility with a smaller overall risk increase
DCT testing - assessing by comparing false positives and true positives
SAme number of true positives as false positives - straight across the board = test has no predicitive value at all
The higher the curve deviates from the middle line
the better the test is
Ideal world in testing
100% true positive rate
More SNPS does not mean
better data
more SNPS don’t mean better data
from 1 to 10 SNPS massively increases the ability to predict a particular trait however after 10 you see a saturation effect and the line flattens off significantly
What should you consider in regard to SNP numbers when running GWAS
- have to weigh up the cost involved of putting extra SNPS on and analysing them between what actual increasing information you’re getting.
- If you add an extra thousand SNPS and you go from a predictive value of 0.68 to 0.69 then you could argue it’s not worth your time
Direct to consumer genetic testing marketed with the promise of providing
- Predictive genetic risk assessment for a variety of health conditions (i.e. diabetes, cancer, obesity)
- Information regarding response to and/or side-effect risk of certain pharmaceuticals (i.e. clopidogrel, statins) (pharmacogenomic information)
- Carrier status of single gene disorders (e.g. cystic fibrosis, Tay Sachs disease) (Important to remember they rarely test for all the variants which means you’re a carrier, so they may only look at only 1 or 2 variants whereas in CF there are over a thousand variants which would mean you’re a carrier)
- Ancestry information and link to relatives
- Medically irrelevant information such as bitter taste perception, ear wax type, curly hair.
DCT tests - important to remember
interpretation is needed
DCT test - what about the results
• The results are not individualised (though they are often marketed as personalised genomics) but they categorise people according to levels of risk so the results are far from individualised results.
DCT tests - role of health professionals
ensure that tests that might be useful clinically are conducted according to clinical standards
• Ensure that the tests aren’t used inefficiently and that we’re not over-testing, over-reacting, over-screening people based on information that might be false/ inaccurate
The associations found in DCT tests are very rarely
deterministic they are increases or decreases in susceptibility they may alongside your environmental and lifestyle choices, have an affect on your health outcomes but they will not be the be all and end all.
Decision support tool for health care professionals
• Idea is that you take the patient through a set of reasons about why they might be concerned – this showed that most routes ending up to the DCT test was probably not an appropriate test to use for their concern
when shouldn’t people use a DCT test
o Genuine genetic concern because of family history, pregnancy issues, carrier status etc – there are much better clinical pathways to get testing and support for those conditions instead of Direct to consumer testing
How do different companies compare?
- different epidemiological data
- different SNPs assayed or included in analysis,
- Different algorithms to combine risks
How do different companies compare? 1. different epidemiological data
= Generating different absolute (background) risk based on their population data set
i.e. one of our risks of obesity was 49% against an average risk of 59% so genetically less likely to be obese than the average person BUT This depends on what you’re considering the average person to be genetically - if you were to look at a Japanese population they’re probably less likely to have a 59% average background risk of obesity compared to the UK or US populations = caution needs to be taken
How do different companies compare? 2. different SNPs assayed or included in analysis,
= different relative (genetic) risk –
different companies use different SNPS in their analysis, either in the ones they assay or the ones they report, some will assay the same SNPS but only report on a different subset
i.e. type 2 diabetes risk – all kind of risk predictions based on 11 SNPS in 23 and me and 21 in deCODEme and 18 SNPS in Navigenics
so depending on which company you go with, you’ll get different result about what your risk of Type 2 diabetes is even though you’re absolute genetic risk for diabetes will be the same (just depends on what the company tells you)
How do different companies compare? 3. Different algorithms to combine risks
= Completely different results!
each company will have a bespoke algorithm to combine these risks and make predictions which can lead to completely different results from being high risk with one company to lower risk in another company.
iThis shows the inherent issues with using these tests for any kind of clinically informed decision making
Concerns with DCT
Validity Clinical Utility Understanding Family and Children Data security Regulation Burden on the NHS
Concerns with DCT - Validity
Analytical validity – did they get the genotyping right? Overwhelmingly for rare variants for monogenic variants they did not get the genotyping right . considering BRCA genes and the potential for therapies and interventions you really want to be correct in your genotyping
Clinical validity – On the assumption that genotyping is correct – can you actually use it? Is there anything you can do with it in the clinic? This will vary depending on the gene and associated condition you’re looking at
Concerns with DCT - Clinical utility
o Related to clinical validity
o Has it got any usefulness in the clinic compared to whether it’s even clinically correct?
Concerns with DCT - Understanding
o Understanding of the individual consumer receiving this information - Almost none of these companies provide any kind of genetic counselling service before or after testing
o Understanding of healthcare professionals – how many of our healthcare professionals have any idea what any of this means? Could they over-interpret? Or will they not understand at all and not take any action on legitimate concerns?
Concerns with DCT - family and children
o When you have a test done you’re not just getting information about yourself you’re also getting information about your family
o Important to have a conversation with family before getting a test
Concerns with DCT - Data security
o Sending your DNA to a company and downloading your results from an e-mail – important to consider data security
Concerns with DCT - Regulation
o Largely unregulated most of the regulation in the UK come from the advertising standards legislation
More about what they say they can do and what they say the tests are good for etc
Lots of disclaimers on the websites “this should not be used for medical decisions” or “this tells you nothing about your risk” when in fact the whole marketing strategy is just that.
Concerns with DCT - Burden on the NHS
o Overwhelmingly people won’t know what to do with this so they turn to their GP. If this information is incorrect in the first place then there’ll be a huge waste of money and time of GPS doing tests on needless people.
Benefits of DCT -
Empowerment, Personal utility, clinical utility, engagement, citizen science, innovation
Benefits of DCT - empowerment
Empowerment, Personal utility, clinical utility
Benefits of DCT - personal utility
o The argument that it doesn’t need to have any kind of clinical validity or utility to be useful to the person. You might be interested in the info even if you can’t use it to affect your health care
Benefits of DCT - Clinical utility
o There are cases of people that have been picked up with conditions because of personal genomic test that they otherwise would not have been picked up for. Potential for clinical utility but only if the analytical validity is high enough.
Benefits of DCT - Engagement
With genetics
With personal health
o The companies can be great in the way they present data – very educational, receive lots of information which is great as the more people get engaged in genetics and their own personal health the better that’s likely to be
Benefits of DCT - Citizen Science
o People wanting to donate their data and information to help science – 23 and me are now a major collaborator with research institutes around the world as they have millions of people’s genetic data and phenotypic data( self-reported) to share so major benefits to crowdsourcing of research
Benefits of DCT - Innovation
o Lots of companies popping up with several tests which generates lots to debate and pushes science forward.
Leigh’s concern with DCT (surprising revelations)
Population level tests not necessarily applicable to the individual
Diagnosing someone with CF due to a pathogenic allele - large clinical implications - lots of extra tests to ensure there hasn’t been a misdiagnosis with the GP - increased burden on NHS
Companies communicate risk - all about interpretation
Making sense of all the data -
• Raw data such as VCFs (for indels in this example) can be annotated with software such as Annovar to provide
o consequence information (exonic or intronic, which gene is it in, is it a frameshift, is it a missense, is it a stop gain, how frequent is it in other populations)
o allele frequencies,
o database IDs and
o clinical predictions.
Caution - police
• killer tracked down because of personal genomic testing
• the police were lost so generated a DNA profile from the crime scene and uploaded it into a website to find relatives – gene matching service
• they found a relative of the killer in the database and tracked down the actual killer and convicted them based on their genetic similarity to a relative who uploaded to the site
• Ethics?
o Is it ethical to use the databases to solve crimes
o Might lead to people being a bit more cautious about sharing in the future
