Wednesday 28 July 2010

Exploring my genome with 23andMe - traits

This is the third in my series of articles on my experiences of testing with 23andMe. I wrote yesterday about my carrier status and drug responses, and on Monday I discussed my disease risks. I'm now going to look at my results for the various traits. The 23andMe test currently covers 40 traits from the mundane, such as eye colour, hair curl and height, to the obscure such as "asparagus metabolite detection" (the ability to detect a distinct smell in your urine after eating asparagus). The screenshot below shows a selection of my results. (Click on the screenshot to enlarge it.)
One of the useful aspects of the data on traits is that you have the opportunity to see how your genetic predictions match the reality. According to my genotype I have a 72% chance of having blue eyes, a 27% chance of having green eyes and a mere 1% chance of having brown eyes. My eyes are actually green so the prediction is accurate, but this is also a useful reminder that the most probable prediction is not necessarily the reality.

For many of the traits 23andMe provides reports on more than one marker. The results can often be contradictory, again reflecting the complex pattern of genetic inheritance. There are for instance two reports on hair curl.  The confirmed research report suggests that I should have "slightly curlier hair on average" whereas according to the preliminary research report I am predicted to have only a typical amount of hair curl. In reality I have a mop of thick curly hair. The reports do explain that "this biological trait has a strong genetic component, although a simply inherited 'hair curl gene' that completely determines hair texture has not been found and likely does not exist. Instead, variations in several genes affecting different aspects of hair development probably combine together to determine the shape of your 'do". (I've never heard the expression 'do before but I presume it is perhaps an abbreviation for hairdo.) My sons have both inherited my thick curly hair so there is a strong possibility that they will also have inherited my marker for male pattern baldness. According to the preliminary report: "Sons of women with this genotype have a 50% chance of inheriting greatly decreased odds of male pattern baldness."

As another example, 23andMe provides preliminary reports on two markers for longevity. I have one marker which gives me higher odds of living to 100 and another marker which gives me typical odds of living to 95. The reported studies were however both very small – a study of 212 Ashkenazi Jews and a study of 615 Japanese men.

I have never been particularly athletic so it is no surprise to learn that I do not have "fast-twitch muscle fibre" and am therefore unlikely to be a fast sprinter. I didn't really need a DNA test to tell me that I have wet earwax, but on reading the report I now have a new word cerumen - the technical term for earwax - to add to my vocabulary!

One of the most interesting findings in my traits reports was the discovery that I have a genotype which makes me resistant to the most common strain of norovirus. Apparently because I have two As on this particular SNP I lack a functioning FUT2 gene. I was fascinated to learn that about 20% of people with European or African ancestry, but very few people with Asian ancestry, share this trait. I'm not however in any great hurry to go on a cruise to test my resistance!
The most innovative feature of the 23andMe service is that customers are given the opportunity to participate in the company's research programme, 23andWe. The company's first scientific paper, Web-Based, Participant-Driven Studies Yield Novel Genetic Associations for Common Traits, was published in June this year in the open-access peer-reviewed journal PLoS Genetics. The data on some of the traits on which the company reports, such as "photic sneeze reaction" and the above-mentioned asparagus metabolite detection, are taken from this paper. While the traits are somewhat trivial and insignificant, the open web-based research model has been validated and there will no doubt be many interesting discoveries in years to come. It is very exciting to be able to engage with your data in such a way and to have the opportunity to find out how your results compare with those of other participants and to see how your data has been used.

One of the best features of the test is the interactive nature of the website and the fact that your results are constantly updated. I've already had a number of new results in the last few weeks, the most recent of which was for leprosy susceptibility. Not all of the information will necessarily have any practical application. I am highly unlikely ever to contract leprosy and probably many of the other diseases for which I have reports. However, new studies are being reported every week and the value of the test will grow over time. Although I was primarily motivated to purchase a 23andMe test for the ancestry aspects I have found the medical aspects very interesting and educational. There is no better way to understand genetics than to have the opportunity to explore your own genome.

See also

Part 1 Disease risks

Part 2 Carrier status and drug responses

Part 4 Ancestry

© Debbie Kennett 2010

Tuesday 27 July 2010

Exploring my genome with 23andMe -
carrier status and drug responses

This is the second in my series of articles on my test with 23andMe. I wrote yesterday about my disease risks. I am now going to look at my carrier status reports and my drug responses. 23andMe provides information on your carrier status for 24 diseases and conditions. Fortunately I tested negative for all 24 diseases. I must confess that I had not heard of the majority of diseases listed. When I explored further it transpired that many of the diseases were quite rare, and a surprisingly large number were particularly prevalent in the Ashkenazi Jewish community (eg, Bloom's syndrome, Canavan disease, mucolipidosis IV and the wonderfully named Maple syrup urine disease type 1B). I don't know whether the SNPs (single-nucleotide polymorphisms) chosen for the 23andMe test are a reflection of 23andMe's customer base or if it is simply that more research has been done into the inheritance of diseases in the Ashkenazi community.

It is important to remember that there are often a large number of SNPs associated with particular diseases and 23andMe only tests a small fraction of these SNPs. It is therefore quite possible to test negative for a particular disease but still to be a carrier.  To its credit, 23andMe does a commendable job of educating its customers. There is an excellent post on their blog, Absence of evidence is not evidence of absence: understanding what genetic testing can (and can’t) tell you, which explains the problem of interpretation very succinctly.

One of the most controversial aspects of the 23andMe test is the inclusion of the SNPs for three BRCA breast cancer mutations. The company make it very clear that "the absence of these mutations does not rule out the possibility that a person may carry another genetic variation that increases the risk of these diseases". (See screenshot below - you will need to click on the image to enlarge it.)
Interestingly again, the SNPs selected account for the majority of breast cancer cases in Ashkenazi Jewish women. Only a tiny proportion (0.2%) of 23andMe customers test positive for the BRCA mutations. Anne Wojcicki's presentation to the Food and Drug Administration last week provides statistics on the prevalence of some of the diseases and conditions in the 23andMe database. 

My only quibble with the presentation of the carrier status data is that you are urged to consult a genetics counsellor if you have any questions about your results. The link provided is to a commercial genetics counselling organisation in America. This might be appropriate for an American customer who presumably has to pay to see a doctor and for whom a genetics counsellor might be a cheaper option.  It is however not much help if you don't live in the US. For customers in the UK it would be more appropriate to make an appointment with a general practitioner for which of course there is no charge.

The drug responses are probably the most useful part of my 23andMe data. I have learnt that my genetics indicate that I might possibly have a reduced response to hepatitis C treatment and an increased sensitivity to warfarin. It is of course quite possible that I will never need either of these treatments but if I should do so then I am sure that the information will be beneficial to my general practitioner who might then be able to target the dosage accordingly and perhaps save the National Health Service some money in the process.
I was intrigued to learn from a preliminary research report that I have a marker which means that I might be a fast metaboliser of caffeine. This would give me a reduced risk of a heart attack from drinking coffee. Although I don't like coffee I do consume large quantities of tea every day!

Of no practical use, but interesting nonetheless, one preliminary research report suggests that I have a genotype which puts me at substantially higher odds of becoming addicted to heroin! The study was carried out on just 139 heroin addicts primarily of Swedish origin and it remains to be seen if these results can be replicated.
Although I've not made any major discoveries about my health it has been very interesting exploring the wealth of data on the 23andMe website. The value of the test will grow over time as more studies are published and the results are incorporated into the 23andMe database.

See also

Part 1 Disease risks

Part 3 Traits

Part 4 Ancestry

© Debbie Kennett 2010

Monday 26 July 2010

Exploring my genome with 23andMe - disease risks

In recent months I have been following with interest the experiences of my friends in the genetic genealogy community who have tested with the personal genomics company 23andMe. The 23andMe test looks at over 500,000 SNPs (pronounced snips) - markers in your DNA known as single-nucleotide polymorphisms - and provides you with information on your ancestry and your predisposition to 163 traits and diseases. The Complete Edition 23andMe test normally costs US $499 (about £321). It is also possible to purchase separately the Ancestry Edition at $399 or the Health Edition at $429. The Complete Edition is however the only 23andMe test which gives you access to your raw data, and is therefore the most useful. Although I have been curious about the process I really could not justify spending nearly £400 on such a test. However, when 23andMe had a one-off sale on DNA Day at the end of April I jumped at the opportunity to purchase the Complete Edition test for just $99 (£63). I had to pay an additional $70 for Fedex delivery. The total cost charged to my credit card was £117. I have had my results for several weeks now but it has taken me a while to explore and digest all the information. I will share my experiences of my 23andMe test in a series of  blog posts, and will focus first of all on my disease risks. I should caution that I have no scientific or medical qualifications. There is a growing body of healthcare professionals and government agencies in America who believe that consumers such as me should not be permitted to purchase these tests without the intermediary of a doctor or genetics counsellor because we are supposedly incapable of understanding the results and, despite lack of evidence, there is a theoretical risk that we might make inappropriate health care decisions based on our findings.

By far the hardest part of the test was grappling with the packaging and then working out how to send the kit back to America. The Fedex instructions were very confusing, and no UK contact details were given. I eventually found a UK telephone number on the Fedex website and was then able to ring up and make arrangements for a courier to come and collect the kit. Before my test could be processed I was required to read and agree to a very lengthy and thorough consent document which reminded me first and foremost that it is "not a test or kit designed to diagnose disease or medical conditions, and it is not intended to be medical advice". The full text of the consent form can be read on the 23andMe website.

The results are all made available through a personal account on the 23andMe website. I have provided below a few sample screenshots. You will need to click on the images to enlarge them. 23andMe distinguishes between "established research" and "preliminary research". For results to be classified as  "established research" (shown on the screenshots with four yellow stars) the association must be widely accepted in the scientific community or validated by at least two studies on more than 750 people with the trait or condition. "Preliminary research" reports are shown with grey stars and are graded depending on the size of the study or studies in question. These findings have not been replicated in other studies and have therefore not yet been confirmed by the scientific community. The results for Parkinson's disease and breast cancer are initially locked, and you are required to read the warnings before deciding whether or not to access these results.
My most elevated risk was for type 2 diabetes, for which 23andMe tell me that I have a 23.6% risk versus an average risk of 18.2%. The screenshot below shows the way that my diabetes risk is presented and explained.
As an illustration of the complexity of so many diseases 23andMe reports on nine markers which have been associated with type 2 diabetes. I have a slightly increased risk on five of those markers but a slightly decreased risk on the remaining four markers. It is however still possible to contract a disease even if you have a reduced risk or, conversely, you might have a higher than normal risk of a disease but never suffer from the condition. For each disease or trait a large amount of information is given which is explained very clearly and presented in a way which makes it very easy to understand. My only criticism is that in the resources section all the links are specific to the US. The majority of 23andMe's customers are in the US so this is understandable, but it is not particularly helpful for those of us who do not live in America. You are also encouraged to consult with a genetic counsellor about your results and a link is provided to a commercial company in America which charges between $99 and $375 for a consultation. Although there is nothing in my results which gives me any cause for concern, if I did have any worries it would have been useful to know who to approach. I'm not aware that we have similar networks of private genetic counsellors in the UK and presumably access to a counsellor would only be possible on the National Health Service with a referral from a general practitioner.

My results also show that I have a slightly elevated risk of age-related macular degeneration and exfoliation glaucoma. I have no family history of either condition and I have never smoked, both of which can be contributory factors.  I have been short-sighted since childhood and, because I wear contact lenses, I go for regular check-ups with my optician. I will be therefore be able to alert my optician to the report, and ask her advice when I next go for a check-up.

Amongst the preliminary research reports I was particularly interested to see that I have a genotype which, if the results of the preliminary study on just one marker can be replicated, would put me at substantially higher odds of contracting diffuse-type stomach cancer. My maternal grandmother died of stomach cancer, albeit at the advanced age of 89, so I quite possibly do have a genetic predisposition. The study cited looked at 925 patients and 1,396 healthy controls from Japan. It will be interesting to see if these results can be replicated in other studies and particularly in European populations. One of the most useful features of the 23andMe service is that your results are continually updated with new findings from published papers.

The remaining results are organised into two further sections - those for which I have a decreased risk and those for which I have a typical risk. The way that the risks are presented is somewhat confusing. According to 23andMe my biggest risk is for obesity. I have a typical risk of 50.6% which is slightly lower than the average of 59%. It seems odd that obesity, my highest risk factor, is not highlighted in red at the top of the page whereas bipolar disease, for which I have only a 0.2% risk versus the average of 0.1% is flagged in red.
I was also very surprised at the figures that 23andMe quotes for obesity. I am advised that on average "59.0 out of 100 women of European ethnicity will get obesity between the ages of 13 and 59". The source for this figure is not given but I rather suspect the figure applies to American women of European ethnicity and not to the European population as a whole, especially as the obesity page is accompanied by a map of America with a breakdown of the obesity levels in each individual state. If so then the percentages for my other disease risks, and especially diabetes which has a strong association with obesity, would need to be suitably adjusted.

I have had no big surprises from the 23andMe test, but it has certainly been an interesting experience exploring my results. I will write more about some of the other features in future posts.
Update 31 July 2010
In today's Times newspaper there was a report from Mark Henderson, the Science Correspondent, on the contradictory population risk data provided by the leading personal genetics companies. If you have a subscription to The Times (you can sign up for a weekly subscription for a nominal fee) you can read the full story here along with the associated commentary from Daniel MacArthur here. Mark had his DNA tested with three different companies: 23andMe, deCODEme and Pathway Genomics. He discusses his results here and in a blog posting hereThe Times story begins: 
Personal genetic testing companies are to change the way that they present health risks after an investigation by The Times exposed how they can mislead people about their chances of developing disease. Three services have agreed to correct problems that can produce confusing and potentially inaccurate personal risk predictions for serious conditions such as heart attacks and diabetes.
The companies are giving the same individuals widely different risk estimates for some diseases because of uncertainty about how widespread the conditions are in the general population. This uncertainty affects the personal risks sent to customers but is not declared in their results... 
All three companies have announced that they will be reviewing their services and have said that "they would provide clear sources for population risk of diseases, and warn customers that the personal risks calculated might not be accurate for everybody." 23andMe have advised that they would "make adjustments within weeks".  I have been very impressed by the commendable way in which the companies have responded to these concerns and I look forward to seeing the amendments to my 23andMe results in due course.

See also

Part 2 Carrier status and drug responses

Part 3 Traits

Part 4 Ancestry

© Debbie Kennett 2010