Personal genetic testing is advancing rapidly. But beware of overselling

GENETIC testing promises a lot. In particular, it promises to tell people things ranging from their risks of developing ailments as diverse as heart disease, cancer and autism to how much coffee they can safely drink. It also promises a lucrative market for those doing the testing. Single-gene tests, such as those for particular forms of genes that predispose people to breast cancer, have been available for a while. This year, however, has seen the arrival of commercial versions of techniques that can sample a person’s entire genetic make-up, and do so in a way that will enable him to benefit from future discoveries as well as existing knowledge.

In many cases, knowing the risk will also allow (and might, indeed, encourage) someone to modify his behaviour to avoid a disease he is at risk of—or, failing that, to mitigate its consequences. Nevertheless, concerns are being raised about the accuracy of some tests now on the market, and also their usefulness when the results are supplied direct to consumers, rather than with professional medical advice.

Last year America’s Government Accountability Office said that genetic tests it had bought from four websites misled consumers into thinking they were at risk of ailments such as heart disease, osteoporosis and type 2 (late onset) diabetes. In some cases, firms tricked consumers into buying “personalised supplements” that were actually no better than ordinary vitamins, but cost $1,200 a year. More recently, on December 4th, the British government’s Human Genetics Commission published a review calling for greater regulation of genetic tests. One member of the commission recently went so far as to brand them a “waste of money”.

Testing times
Fraudsters can, of course, be found in any industry. But another part of the problem has been the science itself. Eric Brunner, an epidemiologist at University College, London, points out that he and his fellow researchers have struggled for years to try to understand the genetic basis of common diseases. The field is plagued by small, weak studies that are hard to turn into statistically robust conclusions. As a consequence, most findings of associations between diseases and particular genes (which often get reported widely in the media when they are announced) do not stand up to later scientific scrutiny. Yet the pace of commercialisation means that companies have often started selling tests based on the earlier studies by the time their results are discredited.

Paul Pharoah, an oncologist at Cambridge University and a critic of some gene-testing firms, says that in the past tests for a gene called SOD2 have been available, with companies claiming it was associated with an increased risk of breast cancer. However, Dr Pharoah and his colleagues recently published the results of a large study which showed that SOD2 is not associated with breast cancer after all.

With luck, this sort of thing will become rarer as scientists gather more data. However, some people worry that even those tests that do what they say on the packet may not actually give the consumer useful information. A possible example of this is a gene called TCF7, which is the subject of a test marketed by deCODE, an Icelandic firm. One form of the gene, TCF7L2, is strongly linked to type 2 diabetes. Having two copies of TCF7L2 (one from each parent) doubles your risk of getting diabetes—that much is well established. But some researchers reckon that if you do not have any of the other risk factors for diabetes, your chance of getting the disease will be so low in the first place that this doubling is not worth knowing about.

The processes that lead to most diseases are not, however, the result of a single genetic failure. Instead, they involve hundreds if not thousands of genes interacting with one another. In the past geneticists have concentrated on genes that have large individual effects when they go wrong, because such effects are easy to spot. But particular combinations of genes that are not individually significant may also be important.

This raises the question of how many genes a test needs to look at to yield a meaningful estimate of risk. Though the answer is not yet clear, Dr Pharoah reckons it is likely to be more than just two or three. Tests that look at only a handful of genes, he thinks, are simply not useful.

The answer to that, of course, is more information. And it is here that the new generation of genome-wide tests comes in. These tests (which are being offered by deCODE and also by 23andMe, a Californian company) analyse the pattern of hundreds of thousands of bits of DNA known as single-nucleotide polymorphisms (SNPs, often pronounced “snips”). A SNP is a point on a chromosome where DNA routinely varies from one person to another. Many SNPs are associated with disease-causing forms of genes, and more associations are being discovered every month. Indeed, this year has seen a bumper harvest of strong, well-replicated associations between SNPs and diseases such as cancer and diabetes.

Analysing SNPs is not the same as analysing a full genetic sequence. For that to happen routinely, the technology will have to improve a lot (see article). However, deCODE and 23andMe are offering individuals a far broader look at their genes than has been possible before—and, in the case of deCODE, are backing their opinions with original research done by the firm, rather than merely drawing on literature published by others.

Yours, for a SNP
Both firms take pains to point out that what they are offering is an “information service”, not a “test”, and certainly nothing that is intended to diagnose a medical condition. And with his firm’s existing test for TCF7L2, Kari Stefansson, deCODE’s boss, seems to agree that testing for this gene should be prescribed “in most instances” by a physician. However, he concedes that the wide-ranging SNP test will also tell customers how many copies of TCF7L2 they have, whether they have been through a doctor or not.

Whatever the disclaimers say (and one purpose they have is to help firms to avoid regulation) people are clearly going to be getting medical information from such tests. For example, overweight customers who find they have two copies of TCF7L2 really do have something to think about.

In time, of course, the regulators will catch up. Indeed, Dr Stefansson hopes that deCODE’s products will eventually become approved diagnostic techniques, rather than mere information services. But for the moment, this is still an area where buyer should definitely beware.

And that is probably right. The technology of testing is improving fast and regulation risks slowing progress. Moreover, physicians’ calls for scrutiny should themselves be scrutinised, because genetic testing inevitably transfers power from doctors to laymen.

That transfer of power brings responsibility, of course—the responsibility of consumers, aided by the gene-testing companies themselves, to interpret their new knowledge sensibly. If they do not, doctors’ surgeries may be flooded with what have come to be known as the worried well, and regulation is sure to follow. If people do take responsibility, however, a healthier life awaits them.

“Do you store my sample?

Once you purchase genotyping services from 23andMe, you will receive a saliva sample collection kit and you will then send your sample directly to our contracted laboratory. The laboratory will extract DNA from your saliva sample and use it for the genotyping analysis. After the analysis is complete and the data have been incorporated into our tools for you to use on our website, all DNA and saliva samples will be destroyed. Neither 23andMe nor its contracted laboratory will save or store any DNA or saliva samples.”

Tracking ancestry and such however, would be quite interesting … although probably more so for Americans – white Britons don’t really have many different places to have come from, its quite predictable really, I could probably gather as much from one’s last name

You pay the $1k to get a return. The return is the “insider” knowledge of what insurance you have to purchase to actually invest “smarter” than all the other insurance’s clients (and therefore to bypass the economic logic of any insurance).

This leads to the following process:

Phase 1:
- only very few people buy the service (i.e. because it’s still expensive)
- they gain an (economic) advantage over the rest of the population

Phase 2:
- other people begin to understand the economic advantage of buying the service
- the quantity of subscribers begins to become “critical”

Phase 3:
- “too many” people use the service and the economic logic of the insurance does no longer work
- insurances will pressure the governments to make the service mandatory for everybody (and the results transparent for the insurances), in order to restore the economic viability of the insurance

Conclusions:

1. companies like 23andMe will have established a devilish business model (sort of state-sponsored) and will eternally make an awful lot of cash

2. everybody will be confronted with their future from the beginning of their lives, if they like it or not

3. companies will therefore have a great incentive to create and sell new solutions which will allow future parents to influence “the outcome” (their children) even before conception, in a way that will produce “better” children (children that will genetically be more perfect and finally cost less)

4. the poorest of the population will have less access to such solutions and the gap between poor and rich will become bigger

=> The mere existence of services like 23andMe is a self-fulfilling prophecy (and therefore a very clever business model) which will change our lives a lot

=> If you choose to use such a service, you agree to potentially be confronted with the worst. Therefore, it’s obvious that you might feel the need to purchase additional insurance coverage after receiving the analysis. This is the reason why you should talk to nobody about you doing the analysis – not even write about it on a public blog (got it, Mike?). This means that the service in its current form is ill-conceived, it should be completely anonymous in order to be “perfect”.

Why do they want to own all their clients’ saliva?

There seems to be no valid reason for that: it almost costs nothing to send them another sample of saliva in the future, should you want to run another test.

If they “own your saliva” it becomes their property and you therefore loose any kind of control over it, meaning:
- they can do whatever they want with it (i.e. 1) extract information you have never requested to be extracted, for whatever use they see fit in order to create profit or 2) forward or even sell it to whom they see fit – unless they somewhere explicitly state the opposite),
- you won’t be able to force them to destroy it,
- you won’t be able to have them send your saliva back to you.

Honestly, I would never blame Mike for not handing them over his saliva. I’d rather see it as a proof for common sense.

John Treadway (JT): “Well, um, er, … yes, that’s my signature.”

ITA: “Thank you. Your honor, I submit into evidence the credit card receipt for $1000.00 paid to 23andMe, a DNA testing service, which is dated 6 months prior to the insured’s application with my client’s company. Further, I submit to you the results of this DNA test which clearly indicate a significantly increased chance of cancer, Alzheimer’s, and hear disease due to obvious genetic flaws.”

JT: “OBJECTION!”

Judge: “Overruled, you pathetic loser… Case decided in favor of Insurer. Next case!!!”