With the highest-profile seller of $99 genetic tests under fire, will public trust in personalised medicine suffer, an ethicist wonders

IT'S 2008. The New Yorker is chronicling a celebrity "spit party", at which notables nicknamed the "Spitterati" eject saliva into tubes to find out their risk of developing illnesses such as diabetes, heart disease and cancer. The firm involved is 23andMe, a direct-to-consumer genetic testing company whose service was named Invention of the Year by Time magazine.

Fast-forward five years. 23andMe receives a demand from the US Food and Drug Administration (FDA) to stop selling its health-related tests pending scientific analysis. In a separate event, a Californian woman, Lisa Casey, files a $5 million class action lawsuit alleging false and misleading advertising. 23andMe suspends sales of its test, putting paid to its target of reaching 1 million customers by the end of 2013. Where did it all go wrong?

In November, after what the FDA describes as years of "diligently working to help [23andMe] comply with regulatory requirements", the agency sent a scathing letter to the firm's CEO Anne Wojcicki. It stated that 23andMe's Personal Genome Service was marketed without approval and broke federal law, since six years after it began selling the kits, the firm still hasn't proved that they work.

Doubts go back a long way. In the year of the spit party, the American Society for Clinical Oncology commissioned a report that concluded the partial type of analysis involved wasn't clinically proven to be effective in cancer care. In 2010 the US Government Accountability Office concluded that "direct-to-consumer genetic tests [involve] misleading test results... further complicated by deceptive marketing".

What 23andMe offered was a $99 test for 250 genetically linked conditions, based on a partial reading of single-nucleotide polymorphisms (SNPs). These are points where the genomes of different individuals vary by a single DNA base pair. There are some 3 billion base pairs in the human genome this test targets only a fraction of them. Different companies sample different SNPs and so return different results for the same person.

To illustrate this point, in his book Experimental Man, science writer David Ewing Duncan recalled how he received three conflicting assessments of heart attack risk from three different companies. The director of one, deCODEme no longer offering such tests telephoned him from Iceland to urge him to start taking cholesterol-lowering statins. Yet the other two tests one from 23andMe, one from Navigenics, which no longer offers consumer tests had rated him at medium or low risk. Given that some statins carry side effects such as muscle weakness, Duncan might have been ill-advised to follow deCODE's urgent advice.

This is the root of the FDA's concerns. In its letter to 23andMe, it raised the risk that customers could get false information that leads to drastic and misguided medical steps. Wojcicki now says: "We want to work with [the FDA], and we will work with them." But is it too little, too late?

And what of the class action lawsuit, brought by Casey after buying a test? It focuses on the test's accuracy but goes further, targeting what Casey's attorney calls "a very thinly disguised way of getting people to pay [23andMe] to build a DNA database".

By asking customers to fill in surveys about health and lifestyle, 23andMe has been creating a valuable "biobank" for patenting purposes and industry collaboration. The firm has always sought customer consent for use of identifiable data and hasn't disguised its aim. "The long game here is not to make money selling kits, although the kits are essential to get the base level data," says 23andMe board member Patrick Chung. "Once you have the data, [23andMe]... becomes the Google of personalised healthcare."

Read the original post:

Testing times for the consumer genetics revolution