Getting personal

When Time magazine drew up the shortlist for its invention of the year award in October 2008, it did not lack suitable candidates. Yet the judges rejected the claims of the Large Hadron Collider and the sexy Tesla Roadster electric car in favour of an idea that has for years been long on promise and short on results: personal genomics.

Time selected the $399 personal genome service of genomics company 23andMe, a Google-backed startup, and the magazine pronounced that genetic tests for all would start “a revolution that will transform not only how we take care of ourselves, but also what we mean by personal information.” The implication is that 23andMe, and competitors such as deCODEme and Navigenics, herald a new era of personalised medicine.

These personal genomics companies scan a portion of your genome from cells provided by spitting into a tube or swabbing your cheek and use the results to assess your risk of a range of diseases and traits. 23andMe looks for 116, from baldness to breast cancer. Yet the worth of examining up to 1m of the 6bn DNA letters in a human genome is uncertain.

First of all, such tests are not directly predictive. They look for genetic variations that raise a person’s risk of developing a condition like Alzheimer’s, but do not necessarily mean that they will get the disease: a combination of other genes and environmental factors matter too. The results can therefore be confusing to the non-expert, and risk creating needless alarm or false reassurance.

The tests cover only a small fraction of the genetic variations that affect risk of disease, and they can be inconsistent too. When the same person takes tests offered by different companies, there is no guarantee that they will come up with the same results: Nic Fleming, a freelance journalist, found that 23andMe gave him a raised risk of age-related macular degeneration (an eye disease), while deCODEme said that his risk was lower than average. As none of these calculations have to be independently verified, it is very much buyer beware.

My own experience with deCODEme reveals further issues. When my genome was scanned 18 months ago, it suggested my risk of a heart attack was 12 per cent above average. Yet when I logged into the deCODEme website to check my details recently, I found that it had risen to 28 per cent above average. This isn’t due to change in my health or behaviour, but simply because the company has updated its algorithms—my risk estimate will change again when more data are added. My profile did reveal one very clear risk worth knowing about: a threefold increase in my chances of contracting glaucoma. But I already knew this because of a family history of the condition, which is still usually a more reliable guide than tests.

So is personal genomics all hype? Each company quotes arresting stories that appear to suggest otherwise. Sergey Brin, the Google entrepreneur whose wife, Anne Wojcicki, co-founded 23andMe, learned that he has a high risk of Parkinson’s disease from its tests. Jeff Gulcher, deCODEme’s chief scientist, discovered that he had an exceptionally high risk of prostate cancer. Further tests revealed he actually had an aggressive prostate tumour that was still treatable. But such cases are unusual. For now, the value lies in satisfying curiosity, not medicine.

This, however, is unlikely to be the case for long. When the first drafts of the human genome were completed in 2001, sequencing a genetic code cost hundreds of millions of dollars. Today, the cost has fallen so rapidly that the company Illumina recently launched a personal sequencing service for $48,000 (covering all 6bn DNA letters, rather than just the 1m or so that genomics companies until now have been scanning). It may cost as little as $1,000 in a year or two. Some observers predict that reading DNA will eventually become so cheap that it is essentially free.

At the same time, a new approach to disease genetics, called the genome-wide association study, has identified hundreds of genetic variations that are linked to common diseases. Initiatives like the 1,000 Genomes Project, which is creating an atlas of these variations, will further increase medicine’s ability to interrogate the genetic code—helping, for instance, to divide diseases such as type 2 diabetes into narrower categories, with different therapies. Many cancer treatments are already tailored to the genetic signatures of tumours, while discoveries about the genetics of drug metabolism will allow doctors to match treatments to genetic test results. The question is not whether these developments start to influence healthcare, but how quickly. And that will be decided largely by how businesses, medical professionals and society at large prepare.

For personal genomics companies, the challenge is to find a viable business model: deCODEme is already in financial difficulty and, despite its media profile, 23andMe refuses to discuss customer margins or profits. As sequencing gets cheaper, it will be difficult to charge much simply for collecting data: the real money will be in interpreting results. Another source of income could be pooling anonymous data about subscribers into databases, for sale to drug companies.

Medicine itself faces different issues, highlighted by a House of Lords report in July. New IT systems and labs are needed to handle genomic information, the effectiveness and cost-effectiveness of genetic tests must be evaluated, and medical education needs reform. At present, British doctors’ genetics training covers rare conditions caused by major mutations, not the fuzzy genetics of common disease. It certainly does not equip GPs to advise confused patients waving printouts of 23andMe scans, or to use genetic profiles to aid prescribing.

For the rest of us, a key question is: who should see our genetic information? A British moratorium on the use of DNA by insurers runs out in 2014. Should it be replaced with a formal ban, so that insurers cannot charge higher premiums to those with a high risk of diabetes or Alzheimers? Or should we trust the market? There is no such thing as a perfect genome, and insurers that take no customers with genetic risks will rapidly go bust.

Genetics is too often considered a deterministic science, revealing traits hardwired by our DNA. This can frighten us when we learn that we carry a gene linked to a dread disease, or breed overconfidence when genetic risks are low. Yet this science more usually works in clues, not certainties, and in concert with the environment. Someone who has a low genetic risk of lung cancer can still raise it substantially by smoking.

Personal genomics will reveal much, but only so much. It should help us to make better lifestyle choices, and help doctors to make better choices about our medical care, but it will not provide an infallible guide to our future health.