23andMe Sends People The Wrong DNA Results [and what is (and is not) interesting about it]

June 9, 2010

As reported Monday (TechCrunch article here), 23andMe recently sent the wrong DNA results to up to 96 customers. Long story short, it appears the lab they farm the sequencing out to (LabCorp) mixed up the labeling on one of their 96-well trays (a standard sized plate) and, well, that meant that when some people logged on to see their results, they ended up seeing the results of someone else.

Beyond the obvious privacy and data security questions this raises, I think it’s the response from 23andMe customers that’s most interesting and illuminating. I will paste their feedback in below since, unless you have an account, you won’t be able to see it (if you do have one, see it here.)

Title: Results in, my son is not my son?

Yesterday I received a message that my son and my daughter’s results were in. I rushed to carrier status and checked, she was negative for all the diseases. I checked my son’s it stated that he was a carrier for hemochromatosis, I was upset. How could he be a carrier and we weren’t. Well my husband’s result’s weren’t in yet so I would wait and see. Still upset I checked family inheritance and noticed my daughter shared with me, and then I checked my son’s. He was not a match for any of us. I checked his haplogroup’s and they were different from ours. I started screaming. A month before my son was born two local hospitals had baby switches. I panicked and I checked over and over. My kid’s were sitting at the computer because we all wanted to see the results. My son laughed but he looked upset. I called my sister in tears. She being the pragmatic one instantly told me to stop crying. She reminded me we took a thousand pictures of his birth and every breath he took the first few months. She told me to check the traits. When I checked for eye color I noticed he was GG (blue eyes) we are AG. My son does not have blue eyes. I compared genes and I noticed he did not compare with any African Americans. His closest results were European. Then I read a post from another person who had wrong results. I realized these results were wrong. I sent an e-mail to 23andme. I received the standard 3-5 days response.Upset, I sent Shirley Wu an e-mail, she said she would forward it to customer service. That’s it. So far I haven’t heard a word from 23andme.
Later I found my son in my bed asleep and hugging my pillow. He did not go to school today, he said he was sick. I told him it’s a mistake.
Someone out there has new results maternal L2c2, paternal E1b1b1a, and those are my son’s results. I looked at relative finder my son has 56 relatives. My son’s Ancestry painting came up today, it says 99% European. I share with 3 of the relatives. I hope they don’t send an invite. Thank god the results were totally off or I would have been in linbo for days. I hesitated before I sent this post, I wanted to give 23andme time to get back to me with some explanation. I am still waiting. I am still screaming inside………

In the comments section another user wrote,

23andme has just sent my son’s girlfriend a notice about a mistake in her processing and took down her results. She was just about to write in and ask how they determined she was Asian. She has spent the day in shock.
I don’t know if she yet called her mother, but I know she was wondering what to ask. She came back 100% Asian. We talked about all possibilities..and some of them were odd and disturbing.
We have spent the entire day discussing her 100% Asian ancestry painting and taking a new look at her pics. Even the high risk breast cancer results were less disturbing to her than what the ancestry meant.
23andme…time for some mea culpas here…

So why does this matter?

Well, before that, I think there’s an important point to be made: the root problem here is not a direct-to-consumer (DTC) specific one - any lab using 96-well plates could have confused labels (or whatever exactly happened here.)

So what’s this “96-well plate” business, you ask? Basically, when a lab is running the same test on a lot of samples, it’s much easier to have one big plate with a lot of wells than it is to have a bunch of individual test tubes. Think about it — would you rather carry around 96 individual test tubes or one plate with 96 wells in it? Since it’s easier and faster, this plate method also contributes to scalability, which then helps bring down costs. So, not only would a lab sequencing for 23andMe use this, but any lab working with a lot of samples that require the same procedures.
Non-DTC examples:
- BRCA1 and BRCA2 genetic testing (which is ordered through a doctor)
- Hepatitis B immunity (checking to see if you have it, that is)

Point being: this mistake didn’t happen because these tests were offered DTC.

Now, what I do think important here is the responses from 23andMe users. DTC companies have long stressed that they aren’t providing medical testing but genetic information (and thus shouldn’t be regulated in certain ways)– but what does all that mean? The first user’s experience (above) shows that the first thing she does when she gets her son’s results is check for carrier status. It’s true that whether or not I’m a carrier for, say, a SNP associated with hemochromatosis, is genetic information — but is getting that information also the result of a medical test? Given the intensity and speed with which she checked for those things, it sounds a lot more like she viewed it as a test than as just “genetic information.” [Maybe I feel this way because “information” sounds so neutral, but “medical test” more closely captures the way we respond to our testing “results.”

To be honest though, I don’t have enough experience with how regulations view the difference to say, but I think the distinction is not so obvious — and the non-obviousness isn’t just a function of how our policies and regulations view the difference. How do people truly view the information they’re given through these tests? An interesting study would be to take people who are told they are a carrier for something and see if they act differently — see if, upon telling their family, if their family treats them differently. How do we process that sort of information?

Why is HIV testing a medical test but the results of a DTC genetic test not? Because an HIV test tells you whether or not you will actually develop certain ailments? The BRCA tests don’t tell you definitely that you will or will not get breast and/or ovarian cancer, but they’re considered medical. Difficult…

Other food for thought:

-The community feedback and advice given to users who wrote about confusing or upsetting results was strong and helpful — I think it shows that, if you’re willing to engage with a community of active and interested participants, you can get constructive information without a doctor middleman. On the flip side, it sounds like users didn’t hear back from 23andMe for days — and thinking you were adopted for a few days doesn’t sound so great.

- DTC genetic testing results aren’t just “for fun.” There were unquestionably traumatic identity crises going on with some of these people. And why? Because when they first clicked through and got their results they were ready to accept those results as telling them deep truths about themselves. They weren’t rapid-fire clicking to carrier status or ethnic background to take a glance, eat some popcorn, and then go out with friends and forget about it. No, this information meant something to people.

It will be interesting to see what comes of this… I’m willing to bet Walgreens is counting its blessings right now, though.

Posted by Heather | Filed Under philosophy, science and tech | 4 Comments 

New FTC Guidelines for Endorsements and Testimonials in Advertising

December 4, 2009

Slightly behind the times but I just had to mention this - on December 1st the new FTC regulations [announcement here, actual guideline legalese here] went into effect.

All in all I think it’s a fantastic move by the FTC. These guidelines, which give interpretational direction regarding the FTC Act, are a big step in the direction of dealing with novel (well, novel since 1980, which is the last time these guidelines were updated) ways endorsements and testimonials are used in advertising.

And I for one am happy to hear there might be an end to the following:

  • those annoying “Lose 50 pounds in a month with SlimQuick” ads where at the bottom, in teeny tiny font, you see “results not typical.”
  • Bloggers who appear to be writing for fun (i.e. people like me) who you later find out were paid to write that glowing review of that new restaurant you were thinking about trying out.
  • Facial potion and lotion ads where you get a “doctor” saying that “clinical trials” show that the product reduces wrinkles in 6 weeks. (Where the “doctor” is really a PhD in creative writing and the clinical trials consisted of a completely shit-ily designed experiment on rats.) 
  • And, yes, “shit-ily” is a technical term.
  • An ad where a doctor says he always recommends X for cholesterol (where it turns out the doctor is a co-founder of the company producing X).

 

Ugh ugh ugh! Even writing out those examples gets me all flustered! It’s so ridiculous I have a hard time understanding why on earth it was allowed to persist for as long as it was. I’m no lawyer (yet!), but I have a seriously difficult time understanding how some of this isn’t tantamount to fraud.

 

But, anyway, a few highlights from the actual guidelines:

Definitions:

  • endorsement = “any advertising message that consumers are likely to believe reflects the opinions, beliefs, findings, or experiences of a party other than the sponsoring advertisers, even if the views expressed by that party are identical to those of the sponsoring advertiser”
  • endorser = “The party whose opinions, beliefs, findings, or experience the message appears to reflect”  [this can be an individual, group, institution, etc.]
  • expert = “individual, group, or institution possessing, as a result of experience, study, or training, knowledge of a particular subject” that’s superior to what your average Joe would know or acquire.

 

More Interesting Stuff:

  • An endorser cannot represent the product in a way that, if the advertiser represented it that way, would be deceptive. (Think of Britney Spears saying “I used SlimQuick and lost 20 pounds in a week! This product is a wonder drug!” where the average weight loss from the product is a pound a month)
  • The following footnote on the whole “results not typical” debacle, “The Commission tested the communication of advertisements containing testimonials that clearly and prominently disclosed either “Results not typical” or the stronger “These testimonials are based on the experiences of a few people and you are not likely to have similar results.” Neither disclosure adequately reduced the communication that the experiences depicted are generally representative. Based upon this research, the Commission believes that similar disclaimers regarding the limited applicability of an endorser’s experience to what consumers may generally expect to achieve are unlikely to be effective…” 
    • Yes!
  • If you lost 100 pounds in 6 months using SlimQuick while also only eating “raw vegetables and exercising vigorously for six hours at the gym” then an ad that says “Betty lost 100 pounds in 6 months on SlimQuick together with diet and exercise” would be misleading because it does not “adequately alert consumers to the truly remarkable circumstances leading to her weight loss".
    • (aka: Betty went nuts, pretty much stopped eating, and dropped a bunch of weight, and 99.99999% of people aren’t going to think that’s what it means to say “lose 100 pounds in 6 months on SlimQuick.”)
  • When a connection between the endorser and advertised product isn’t obvious and might materially affect the credibility of that endorsement, it has to be disclosed.
    • Think of a blogger where you’d never know she was being sent the products to review or a celebrity on a talk show who casually mentions how much she loves Soda X, where she gets paid every time she mentions the product during publicity events.)
    • If you are getting paid to tweet or post status updates on Facebook and the like, you’ve got to disclose it.
    • If you write on a message board about a topic but you are employed by a company who makes that product (or is a competitor of that product, etc.), you’ve got to disclose that to.
      • I wonder if that means both if you are writing in your capacity as an employee or agent of that company and if you’re writing on your own free time.
  • How you are paid to endorse something matters
    • The credibility of your endorsement might change based on if you’re paid one lump sum to endorse the product, if you’re given 10% of all gross profits off that product for the next year, or if you are a shareholder in that company and, accordingly, consumers should be given that information.

 

Short version: “don’t go around trying to deceive people with wild claims or by trying to make it look like someone is endorsing your product out of love (and not money) – it’s messed up.”

Shorter version: “don’t be a slime ball”

Sounds good to me!

Posted by Heather | Filed Under philosophy, science and tech | 4 Comments 

New Doodles and Zittrain’s “Minds for Sale” part one!

November 19, 2009

Unhappy-girl I’ve added two more doodles to the Artsy Stuff > Computer Doodles section. I’ll add them here, too!

I’m also hoping to send out my write up about the really interesting talk I went to last night. Jonathan Zittrain from Harvard’s Berkman Center gave his “Minds for Sale” talk. Info here.

Long story short – I think most of the problems he expressed aren’t unique to cloud computing (or technology at all). Generally speaking, there’s a danger when people agree to do work when they don’t understand the overarching goal of the project or even who they’re working for.

Simple example: Someone puts an ad in the paper: “Will hire someone to flier cars in x shopping center.” You know that’s a heavily Persian area and so you aren’t surprised that the ads are in Farsi. You accept the job, get your $20 and continue on your merry way.

Turns out those little fliers were actually filled with hate speech, threatening to kill Muslims and Muslim supporters.

You didn’t really care (or perhaps better said, you didn’t really think to care) when you took the job what the flier said. And, especially because it was in Farsi, you didn’t have ready access to figuring it out for yourself. But, if the ad would have said “Will pay someone to distribute Muslim hate speech on cars!” you certainly would have declined.

awesome-guy-with-dorky-glasses Now taking that to the land “o” the Internets, things like Amazon’s Mechanical Turk allows people to post simple jobs (like, “what words best describe this picture”) for something like $.01/job. The “turks” have no idea what the larger scope of the project might be and it’s possible if they did, they’d never support it.

So the worry seems to be that things like Mechanical Turk at least have the potential to make people complicit in activities they never would have supported otherwise. But, because you’re just turning one little cog, you might be.

Food for thought…

Posted by Heather | Filed Under Conferences, philosophy, science and tech | Leave a Comment 

My little animation

October 6, 2009

I’m at the Adobe MAX 2009 conference and had the chance to make this little guy in my character animation in flash and illustrator hour session. It’s amazing how a few little tricks make the character come to life!

 

Posted by Heather | Filed Under science and tech | 1 Comment 

Direct-to-Consumer Genetic Testing: A historical and ethical examination

September 8, 2009

[Here’s the paper I wrote a while back. It’s a work in progress but may as well get it out there in the hopes that someone will read it and provide some feedback!]

Introduction and Overview

The goals of this paper are twofold:

First, to place the direct-to-consumer genetic testing of today, with special emphasis on 23andMe, in some historical context. 23andMe was founded in 2006 and while it and its competitors are attempting to change the way people understand their relationship to their DNA, they are not an anomaly but the progeny of both the technological and genetic advancements, and transformed cultural understanding of those advancements, that lay at its foundation. To show this I will spend some time going over what I take to be a few of the major historical milestones that make 23andMe possible. In particular, there are two stories from the history of genetic testing that embody many of the hopes and worries people on all sides of the spectrum associate with 23andMe. After I go over these two stories in some detail, I’ll also briefly mention some other milestones that, if I had more space, also merit a closer look.

Second, to discuss where I think 23andMe ought to be going in the next few years. This will be based on an examination of what I take to be the most important ethical questions direct-to-consumer genetic testing services like 23andMe raise.

A quick word about the ethical framework this paper is using. I am thinking of ethics in the Aristotelian sense. The over-arching question I am primarily concerned with is: what does it take to live a flourishing life? While we may all have our own individual conceptions of what a flourishing life entails, I take it as fairly noncontroversial to say that the freer we are to make autonomous decisions about how we want to go about pursuing a flourishing life, the better. In other words, in order for me to be the sort of person who can lead a flourishing life, I need to be in a position to create a rich picture of what I think that flourishing life entails and then have the ability to make meaningful decisions that will help me make manifest that picture. Direct-to-consumer genetic testing services like 23andMe have the ability to fundamentally impact both an individual’s ability to make meaningful decisions about how to lead their lives and what decisions they think it best to make. Accordingly, it’s important for us to evaluate what the fallout of that is if we hope to continue to put ourselves in the position to lead good lives.

Historical Context

Tay-Sach’s Disease: the power of cheap technology and grassroots efforts

The story of Tay-Sach’s (TSD), a fatal autosomal recessive genetic disorder that Ashkenazi Jews are at a significantly increased risk of carrying, stands as an exemplar of what direct-to-at-risk-community promotion of carrier screenings can accomplish.

In the early 1970s, about 85 years after the disease was initially characterized by Warren Tay and Bernard Sachs, it became possible through the advent of cheap assays to screen large at-risk communities for TSD carrier status. The first community-level heterozygote screening took place in 1971 in both Baltimore, Maryland and Washington D.C. and, since the launch of those community-based screenings and countless others like them, there’s been over a 90% reduction of TSD in North America.[i]

What allowed for the 90% reduction of TSD in North America was not just the availability of cheap assays but the cultural and community initiatives that promoted those screenings. In particular, the Jewish community worked hard to promote these screenings to its youth through organizations like Dor Yeshorim, which, founded in 1983, would go to private high schools with high Orthodox Jew populations and explain that through a blood test students could learn if they were a carrier for TSD. After testing the students, they would give each of the screened individuals a six-digit identification number. If a girl and boy were interested in dating they were asked to call the hotline with their ID numbers. Upon calling they’d be told that they either were or were not compatible based on if the match would result in the possibility of a child being born with the disease. Genetic counseling was then made available.[ii] It’s important to note that in Orthodox Judaism, abortion is considered immoral and the moral status of in-vitro fertilization isn’t clear. The only morally unambiguous way to stop the disease is to ensure that couples aren’t even at risk of having babies who could have the disease, which means two carriers have to avoid having children together since one in four of their children is likely to have TSD.

Other than at the high school level, efforts were and are made to publicize and encourage screenings through things like announcements in at-risk community newspapers, free screenings on college campuses, and presentations at Jewish-youth organizations like Hillel.[iii]

Through the continuing story of TSD we see two important precursors to 23andMe: first, the use of widespread genetic testing as a preventative measure and second, the direct-to-individual promotion of those tests. In the Ashkenazi population in particular, determining one’s carrier status was pushed as being one’s moral duty since the only acceptable way to prevent the disease was to dissuade two carriers from having children. While resulting in less extreme lifestyle changes than mate selection, 23andMe feeds into our desire for information we can use to plan ahead for the best possible future for ourselves and our children. If we know at 30 that we have an increased risk of Alzheimer’s or Type II diabetes, there are lifestyle changes we can make that will help lessen our overall risk. From the TSD story we might say that the more information we have about our genetics, the better position we place ourselves for making decisions that will lead to happy healthy futures.

However, the direct-to-consumer genetic testing of today is also different from TSD screenings in important ways. For one, the genetic basis for developing TSD is known. We know what it means to be a carrier and what it takes to develop the disease. TSD is recessive autosomal and fatal - if two carriers have a child there’s a one in four chance that child will have and subsequently die from TSD. 23andMe, as will be discussed in more detail later, shows people their genetic likelihood to develop diseases and conditions where only tenuous research has been done linking a certain SNP (single nucleotide polymorphism) to an increased or decreased risk of developing a certain condition.

Myriad Genetics and BRCA Screening: direct-to-consumer advertising of genetic tests

On September 12, 2002 Myriad Genetics began a first of its kind advertising campaign to promote their “BRACAnalysis” test. The test looks at what variation of two genes, BRCA1 and 2 an individual has. Depending on which variation or possible mutations a woman has, she will have a 35 to 84 percent chance of developing breast cancer by age 70 and a 10 to 50 percent chance of developing ovarian cancer – both much higher than the risk for the general population.

After the first five month test in Denver and Atlanta the company claimed there were 38 times more calls to its toll-free number in those cities versus control cities and 30% more tests being performed. However, due in part to studies showing that the increased referrals to genetic counselors weren’t only for women likely to benefit and because of a survey conducted by the Center for Disease Control and Prevention that found many doctors in Atlanta and Denver to have an incorrect or at least incomplete understanding of the genetics, Myriad decided to wait five years to start its new campaign in the Northeast United States in 2007. [iv]

Myriad’s direct-to-consumer advertising raised controversy in ways the Tay-Say’s screenings did not because, among other things, many in the medical field felt widespread BRCA testing was unnecessary and unhelpful. According to the New York Times, BRCA mutations account for less than 10% of all breast cancer cases and only 1 in about 400 women have the mutation. Further, a woman with a family history of breast cancer would have a better understanding of her own risk if she instead got the affected relatives tested first. Of the 10% of women believed to develop breast or ovarian cancer due to hereditary factors, not all of them will inherit it through a BRCA mutation. So, to know if the type of inherited breast or ovarian cancer in your family is due to a BRCA mutation, it’s much more useful to have the afflicted family member tested for it. [v][vi]

Myriad Genetics, some argued, wasn’t explaining that important piece of information and instead was promoting the upwards of $5,000 test to women who were just worried about developing breast cancer based on seeing that someone else in their family had.

Further, both positive and negative results for a mutation don’t lead to a clear course of action. When women are told they have a mutation in BRCA2, one course of action is to have a prophylactic oophorectomy. With other mutations, a double mastectomy. These surgeries have the potential to dramatically alter a women’s identity – her body image, her understanding of herself as a women, and her health – and accordingly, should only be undergone after serious consideration and consultation with a genetic counselor. But the risk of women wanting to take drastic measures and doctors being ill-equipped to help explain the pros and cons, make the risk of unnecessary surgeries real.

On the flip side, a negative result may only mean their hereditary risk comes from a mutation other than BRCA. But, since knowledge of other forms of inherited breast cancer may be unfamiliar to patient and doctor alike, prudent precautions, like taking Tamoxifen, an FDA approved drug used to prevent the development of breast cancer in women at high risk[vii][viii], may be improperly dismissed. As Dr. Joanne Armstrong, assistant professor of obstetrics and gynecology at Baylor College of Medicine said in an interview,

“When people receive a small piece of genetic information – disconnected from the whole genetic picture and their environmental risks, they can be misled to believe that they are doomed to develop a disease. That genetic information could be harmful in that it causes a lifetime of unnecessary anxiety. The flip side is that the absence of a certain SNP may be taken for immunity from disease, which could provide false reassurance and encourage people to discontinue preventative measures and screening practices for certain diseases.”[ix]

Myriad Genetics’ BRACAnalysis campaign was the first instance of for-profit direct-to-consumer genetic testing and in so being, blazed a direct-to-consumer path for future companies. Like 23andMe, Myriad is a company looking to make profit and, the more people who find genetic testing relevant to their own lives, the more likely they are to achieve that goal. However, 23andMe, to the best of my knowledge, has not launched widespread advertising campaigns. The vast majority of their publicity comes in the form of news articles and blogs written about them. They host swanky spit parties where big name celebrities[x] spit in a test tube that will then be sent to LabCorps, the company that now does the SNP sequencing for 23andMe[xi], for testing. In sum the parties, beyond being mentioned in the New York Times fashion section,[xii] create an environment where genetic screening isn’t based so much on fear but instead on fun and a desire to social network in a new way. As Anne Wojcicki, one of 23andMe’s cofounders, said in an interview, ““If you want to have a community around psoriasis, we’d like to be able to allow you to form a psoriasis-specific community.”[xiii]

Of course it’s also important to note that 23andMe and its competitors don’t test for the BRCA mutation because, among other reasons, Myriad continues to have a patent on those genes and obtaining a license to perform the test would be prohibitively costly.[xiv] And, even though Myriad was promoting the test directly to consumers, federal law required doctors to actually order the test. 23andMe promotes direct-to-consumer genetic testing and doctors don’t order the test – consumers order the kits directly online.

Myriad and 23andMe differ in important ways but historically Myriad was the first company to realize that marketing genetic testing directly to doctors would be less effective than going directly to consumers. Myriad gave consumers some information through its campaign and believed that information was enough to get them to ask the right questions. And like Myriad, 23andMe shares the belief that with enough online educational tools and resources, consumers can come to understand their genetic code on their own, without the intervention of a doctor. But worry remains that the lesson to learn from the Myriad story is that when companies are aiming to make a profit on genetic testing, they will push tests as being more helpful and relevant to consumers than they really are, which will lead not only to consumers wasting their money (that’s really nothing new) but drawing incorrect and potentially harmful conclusions from the information.

Other milestones worth mentioning

Because the length of this paper is so short, I don’t have enough space to go into depth on all the advancements and changes that have paved the way for the kind of direct-to-consumer genetic testing we see in companies like 23andMe. However, I’d like to at least gesture at a few more that Brian Naughton PhD, 23andMe’s founding scientist, Linda McCabe PhD, and Stephen Cederbaum MD mentioned through my interviews.

From Brian Naughton[xv]:

  • The development of the first microarray by Affymetrix, published in 1991. Microarrays can detect gene expression and SNPs and over time have gotten much cheaper and more accurate, making companies like 23andMe possible.
  • The development of the Illumina 550K genome-wide SNP chip in 2006. Illumina was the first lab 23andMe sent its samples to for testing.
  • The publication by the Wellcome Trust Case Control Consortium (WTCCC) of their first large-scale genome-wide association studies in 2007.[xvi] The project that, according to Brian, “really pushed research forward, and gave us a ton of information on what many SNPs do.”

From Linda McCabe:

  • Newborn screenings – the desire to know what we have before we have it, with the hope that there’s something we can do about it.

From Stephen Cederbaum[xvii]:

  • Pre-natal diagnosis
  • Expanded newborn screenings. Though the law requires certain diseases to be tested for in different states, NeoGen labs is a company that for about $75 will run a much more robust panel than law requires.
  • The cord blood story. A commercial enterprise being sold on the premise that you may never know when you may need it (e.g. if your kid gets leukemia or some sort of procedure to wipe out their own bone marrow), the cord blood story is another instance of us wanting as many resources at our disposal to best prepare for an unknown future.

Other advancements and cultural changes worth thinking about include:

  • The rise of social networking and Web 2.0.
  • The Human Genome Project and increased public awareness of genetics.
  • The increased trust, or at least tolerance, users have when it comes to sharing large quantities of personally identifiable information with private companies. The development of that trust from the start of widespread Internet use, with focus on companies like Google and Paypal, would be worth exploring.

23andMe sits at the intersection of exciting changes on many different technological and cultural fronts. By taking time to think about those connections, we can also think more deeply about how the lessons from those historical events can be applied to 23andMe and the future of direct-to-consumer genetic testing.

The Future of 23andMe: The ethical worries

In a recent article in the New York Times[xviii], Steven Pinker, a well known experimental psychologist and writer, described his 23andMe experience. According to the analysis provided by 23andMe, he had a 12.6% chance of developing prostate cancer before the age of 80, whereas the average white male is at a 17.8% risk. At first he was (unsurprisingly) quite pleased. Then he decided to look at the Personal Genome Project’s [xix] (PGP) data and saw that according to their studies he in fact had an elevated risk of developing prostate cancer. So which is it? Whose data is accurate? What is he supposed to do with the information? What if he never looked beyond what 23andMe told him?

Among geneticists, the answer to the question of whose data is accurate seems to be that they are all right and all wrong… sort of. The human genome is comprised of approximately three billion base pairs and 99.7% of those pairs are totally identical between us all. What’s left are ten million variations that we call “SNPs”. If a company is testing for 500,000 SNPs, like 23andMe, that means 95% of the total variation between people isn’t even touched. Of the ten million variations between us, it might very well be the case that within the 500,000 23andMe tests for, there’s a SNP correlated with a decreased risk of prostate cancer. But of the 9.5 million variations left untouched, it’s more than likely that some of them will bear on that person’s cancer risk, too. The question then becomes: is telling consumers that they are at a 5.2% decreased risk for prostate cancer compared to the average white male at all meaningful? Is it misleading?

To resolve many of the issues raised in the article, companies could start performing full-genome sequences. But that’s not the only issue. In another recent article in BusinessWeek[xx], Dr. Joseph Holoshitz, professor of medicine at the University of Michigan, discussed rheumatoid arthritis. 90% of people with rheumatoid arthritis have a particular SNP. Take a person with the SNP, the disease, and an identical twin. Does the twin have rheumatoid arthritis, too? He has the exact DNA as his brother. The answer: no. Only about 10-15% of the twins will have the disease. But why? The answer lays in epigenetics, to changes in the expression of genes based on causes other than the DNA itself. In other words, it doesn’t just matter what DNA you have “on paper” – it also matters which pieces of that code become expressed. Identical twins have the same underlying DNA and yet just one can be schizophrenic. Just one can be homosexual. And just one may have rheumatoid arthritis. Right now we know very little about how epigenetics work and that raises additional questions about the meaning of the statistics 23andMe and others present to users regarding their risk.

The significance of how users interpret the statistical information companies like 23andMe present them is important for a few reasons. Like Steven Pinker points out, we have a certain essentialist tendency. We want to know “what we really are deep down.” Carl Elliott hits on this when he talks about the distinctly American strand of thought that says to “be authentic.”[xxi] We want to figure out what we are at the bottom and then be true to it. DNA seems to be where many of us think we’ll “unlock” and “decode” the answers, finally able to grasp our essential nature. Because of this, when we’re told we have an increased or decreased risk based on our DNA, we, even when we know there’s much more research to be done, may have a hard time viewing that information in a reasonable light.

In my interview with Dr. Steve Cederbaum, a clinical genetics specialist at UCLA, I asked for his thoughts on the ethical implications of direct-to-consumer genetic testing given where the science is today. His response at first was, “faulty and undesirable practices [like 23andMe] aren’t unethical, they’re just bad medicine.” While I think a large portion of the scientific community views science and science-based companies in this way, I think a grave mistake is being made in separating science and technology from ethics. When I asked 23andme users why they signed up for the service I received responses like, “I wanted to know my genome,” “[I] find the technology fascinating and was very curious to find out more about my medical history and genetic ancestry,” and “I wanted to know my APOE4 status… I was also curious about a number of other genetic traits, such as the fast-twitch muscle gene, bitter tasting, and alcohol dependency.” These users, all bright early adopters, are reading a lot of meaning into the information 23andMe presents. Many users seek the information because they think it will provide them with action guidance, with meaningful information that, when acted upon, will lead to a better healthier life. This becomes an ethical issue if and when individuals or companies present information in such a way as to mislead others, thereby depriving them of the ability to make meaningful decisions. In order for us to pursue flourishing lives, we need autonomy.

For instance, say I’m a vegetarian for ethical reasons. You cook a tasty stew and ask if I’d like to have some. I say “I’d love to – but only if there’s no meat in there!” Now let’s say there is meat in the stew, you know I’m a vegetarian, and yet you tell me there isn’t any meat in there. I then gladly accept some. By lying to me you’ve deprived me of the ability to act in the way I think most aligned with what a good life entails. I find eating meat unethical, I want to live the best life possible, so I don’t eat meat. When you step in and give me false or misleading information (say you said “do you see meat in there? It’s full of vegetables!” instead) you have harmed me – you’ve abused the trust I’ve placed in you and you’ve hindered my ability to live the life I think best. Direct-to-consumer genetic testing services may raise a somewhat analogous situation. As a company they ask consumers directly if they’d like to know more about their genome and then have a mission statement, like 23andMe’s, that states that they want to be “the world’s trusted source of personal genetic information.” Say these companies know what sort of clout people associate with science and genetics. Consumers then buy the kit, give in their DNA sample, and eagerly await the moment they can sign in and discover their fundamental nature. A user sees that they have a 35% higher risk of developing type II diabetes than the average white woman. But on that page the user is given this information[xxii],

“The heritability of type 2 diabetes is estimated to be 26%. This means that environmental factors contribute more to differences in risk for this condition than genetic factors. Genetic factors that play a role in type 2 diabetes include both unknown factors and known factors such as the SNPs we describe here. Environmental factors include obesity, gestational diabetes, giving birth to at least one baby weighing nine pounds or more, high blood pressure, abnormal cholesterol levels, physical inactivity, polycystic ovarian syndrome, other clinical conditions associated with insulin resistance, a history of impaired glucose tolerance or impaired fasting glucose, and a history of cardiovascular disease.”

So what does that all mean for the user? A 35% increased genetic risk but the genetic risk only plays a 26% role – and even then there are unknown genetic factors, too. I think the answer, if there is one, is unclear at best. Is the deep-set belief that 23andMe and others will be showing them the truth about themselves tempered by disclaimers like this? Is it too late at this point? Are the expectations too high? If someone thinks you’re God and you know it and then you tell them they have a 35% higher genetic risk than the average person of developing a disease, though genetic risk is only estimated at 26% are you in a sense taking advantage of their false belief? If Myriad knew that 80% of the women who would get the BRACanalysis test done would do so because of an unrealistic expectation that the test would definitively tell them their risk for heritable breast cancer because of misunderstandings promulgated through their advertising campaign, what should we do about that? Is 23andMe any different?

I asked 23andMe what they thought should be done about this and, through Rachel Cohen, said,

“We have discovered that a number of people do not have a strong understanding of probability - that having a higher than average risk for a disease does not mean they are going to get it.  We make every possible effort to educate our customers about the meaning of genetic risk.  We have found that once risk is explained, people can understand it and use their genetic information responsibly.”

I’ve never had a personality test performed on me but don’t need one to say I lean much further libertarian than authoritarian. And much aligned with my libertarian-leaning “roots,” I tend to agree with 23andMe that education by companies to consumers is an important piece of the puzzle. In the next five years the cost of full-genome sequencing will drop enormously, making it feasible for the vast majority of Americans (the price for 23andMe has already dropped from $1000 to $399 after only a few years). There will be a push, as Obama has already mentioned, to digitize people’s health records. Services like Google health will allow users to put their family history, health records, and genetic information in one place. Personalized medicine will slowly develop and services like 23andMe will help lead the way. And, in many cases, that’s a great thing. Knowing that certain antidepressants work better than others, that I should demand more pap smears because I’m at an increased risk for ovarian cancer, and that I am a carrier for cystic fibrosis can make my life radically better by empowering me to take informed actions. The difficulty lies in the time before that moment of robust knowledge and understanding has been reached. We need early adopters to blaze the way but there’s a very real risk of depriving them of the ability to make meaningfully autonomous decisions if we trivialize or brush to the side the immense power people see in genetics; if we dismiss our essentialist tendencies.
Bibliography

Interviews

  • Linda McCabe, Adjunct Professor, Department of Human Genetics and Pediatrics, UCLA School of Medicine. Interviewed via email between January 4th 2009 to January 11th.
  • Brian Naughton PhD, via email with responses sent back January 8th, 2009.
  • Interviewed by phone January 8th 2009.
  • Stephen Cederbaum MD, via phone on January 8th, 2009
  • Sean McBride, Google user experience designer, interviewed via email with response sent back January 8th, 2009.
  • A survey sent to Googlers and left up on my personal blog. As of January 12th 2009 I received 13 responses. This included
    • 8 23andMe users
    • 5 non-users including;
  • Greg Lennon, founder of SNPedia
  • Direct email responses to my survey questions from
    • Aaron Meyer, a Bioengineering senior at UCLA, and
    • Kenneth Hurst, a Molecular, Cell, and Developmental senior at UCLA and a student research in the UCLA Stem Cell Institute.
  • Emailed with Owen Thomas, a Gawker reporter whose written on 23andMe
  • Attended the Consumer Genomics Law and Policy: Panel Discussion at Stanford on November 10, 2008
  • Contacted but did not hear back from
    • Christina Palmer, genetic counselor

[i] Tay Sachs Disease. Robert Desnick and Michael Kaback. Published by Academic Press, 2001. Pages 1-5

[ii] “Nightmare or the Dream of a New Era in Genetics?” New York Times. Published December 7, 1993. Gina Kolata.

[iii] Conversation with Linda McCabe, Adjunct Professor, Department of Human Genetics and Pediatrics, UCLA School of Medicine.

[iv] “A Genetic Test that very few need, marketed to the masses.” New York Times. Published September 11, 2007. Andrew Pollack.

[v] “Direct-to-Consumer Genetic Testing: Access and Marketing.” Genetics in Medicine. L.L. McCabe and E.R.B. McCabe: 6:58-59, 2004.

[vi] “Ad campaign fuels debate on breast-cancer gene test.” Wall Street Journal. September 11, 2007. Marilyn Chase.

[vii] “Tamoxifen information.” FDA Center for Drug Evaluation an d Research. Last updated July 7, 2005.

[viii] “Tamoxifen” Wikipedia

[ix] “No one really knows what most genetic tests mean.” Better Health blog. October 26, 2008. Dr. Val Jones.

[x] Pictures from one such party can be found here

[xi] Originally Illumina was the company that did 23andMe’s sequencing but they decided to move to a lab certified as “high complexity” under the Clinical Laboratory Improvement Amendments. 23andMe’s explanation can be found here.

[xii] “When in doubt, spit it out.” New York Times. September 12, 2008. Allen Salkin.

[xiii] Ibid

[xiv] “Navigenetics #4 – low penetrance genes v. high penetrance genes.” Genetics and Health blog. April 9th 2008.

[xv] Interviewed via email with responses sent back January 8th, 2009.

[xvi] Wellcome Trust Case Control Consortium website

[xvii] Interviewed by phone January 8th 2009.

[xviii] “My Genome, Myself.” New York Times Magazine. January 7, 2009. Steven Pinker

[xix] The Personal Genome Project is sequencing volunteers’ entire genomes and collecting large amounts of data about what sorts of traits, diseases, and conditions they have with the hope of making all of that information freely available.

[xx] “What gene tests can really tell you.” BusinessWeek. October 23, 2008. John Carey.

[xxi] A Philosophical Disease: Bioethics, Culture and Identity. Carl Elliott. Routledge Press, 1999.

[xxii] This is the exact language presented to users, given to me by Rachel Cohen, 23andMe’s communication manager.

Posted by Heather | Filed Under philosophy, science and tech | 1 Comment 

Next Page »