23andMe, a California based startup co-founded by Anne
Wojcicki, offers a direct-to-consumer Saliva Collection Kit and Personal Genome
Service (PGS), which since its launch in 2007 has genotyped approximately
500,000 individuals. To be clear,
23andMe does not sequence complete genomes, but rather, offers a SNP
(single-nucleotide polymorphism) DNA test that instead records what are
arguably the most important 1 million “snips.”
And while amounting to less than a thousandth of a percent of the total
genome, these snips have been proven to be very informative in describing one’s
unique risk of genetic disease, drug response, personal traits, and ancestral
information. Such insights are supported
by a strong body of scientific literature, which the scientists and
physician-scientists at 23andMe use to interpret raw data (all those SNP’s)
into clean, protected, and user-friendly reports on health and ancestry.
However, in November of 2013 the Food and Drug
Administration (FDA) warned 23andMe to stop offering its primary service of
providing users with a health report on 254 genetically related diseases and
conditions. Following the argument that PGS is a “medical
device” without government approval, the FDA goes on to describe “the potential health consequences that
could result from false positive or false negative assessments for high-risk
indications.” For example, if a woman
sees that she is at risk for breast cancer based on an unfavorable mutation in
her BRCA gene (a gene associated with breast cancer) detected by the 23andMe
assay she may be inclined to undergo prophylactic surgery, chemoprevention, or
intensive screening. With that said,
according to the FDA, it is possible that this woman does not actually have nor
will develop breast cancer— a false positive.
Fearing that customers could make rash decisions based on their 23andMe
health report and thus pursue such expensive and potentially dangerous
treatments, the FDA decided to ban 23andMe’s flagship service. And so as of February 2014, 23andMe is able
to offer ancestry-related reports and raw genetic data, but unable to offer a
health report.
Personally, I think there are reasonable causes for
public examination and regulation of 23andMe, as there should be with any
company diving into an unfamiliar and uncharted market, however promising or
exciting as their service may be. At the
same time, I believe that the FDA is unfairly treating PGS as a medical device
and unwisely pursuing a legal ban.
To expound on the first point, 23andMe is an
information driven company collecting particularly sensitive and complex
information. And like Google, it is
likely that the value of this company is not based upon the service it offers
to any particular end user, but rather the information it is able to gather in
aggregate. So while collecting DNA from
all over the United States could bring us new insights fueling our efforts to
advance science research (population genetics, drug discovery) the sale of such
information could be equally dangerous.
Furthermore, such a service presents an important
ethical question: are we entitled to the information encoded in our DNA? Technically, DNA is what we have inherited
from our parents and it is something we share to a varying degree with our
siblings, cousins, and relatives. As
noted by Charles Seife in an article published by Scientific American, using 23andMe may feel like a personal
decision, but in a sense, it also grants investigators the implicit consent of
family members of generations past, present, and future.
Still, it is not necessary that in regulating these
issues (among others), that the FDA has to ban this service. First and foremost, treating PGS as an
untested medical device is unfair and implies very little respect for the
American public and practicing physicians.
Returning to the hypothetical false-positive, it is worth distinguishing
between phenotype and genotype. On the
genetic level, it is highly unlikely that 23andMe calls a false-positive in the
sense that the sequencing assay falsely calls the wrong mutation at the
particular SNP for the BRCA gene. Such
sequencing technologies are designed with accuracy as the chief concern, and
thus built with in an error rate of around 1 in 7,000— this would easily make
PGS to be the most accurate “medical technology.” So perhaps, the FDA was referring to a
false-positive in the sense that the genotype is right but the phenotype is
wrong. Well, 23andMe is by no means
claiming that a user has breast cancer, but instead, just noting a level of
risk. It is highly unlikely that a
doctor in the United States would perform any serious surgery or complicated
screening on the basis of 23andMe alone.
Last, by banning the service, the FDA has inadvertently
created a void that will be filled by uncontrollable, smaller parties. At the moment, users of 23andMe can access
their raw data, but the experts at 23andMe cannot interpret this data into a
comprehensive health report.
Consequently, users may be tempted to interpret their data for
themselves based upon scientific literature published online. Perhaps there will even be a demand for
third-party businesses to form, specifically designed to interpret raw data
reports from 23andMe. As more of these
situations arise, the FDA will have increasingly less control in matters of
regulation.
As mentioned earlier, there are compelling issues yet
to be addressed regarding the potential misuse of genetic data, but my hope is
that sooner, rather than later, the FDA revokes its ban. Given a more friendly relationship between
both parties, I am optimistic that they can work together to set a historical
precedent, marking the early beginning of consumer genetic testing.