A Letter I Will Probably Send To The FDA

I predict that the current spat between 23andMe and the FDA will be less apocalyptic than some people fear. In particular, I expect 23andMe will submit a lot of paperwork, change some of their wording around, advertise a little less aggressively, stop listing some or all of their “Health Risks” and “Drug Interactions” section, and then be back on their feet in a few months.

But I’m not sure. And even that seems like a bad thing. Although I realize this is probably a very complicated legal game here, I’d like to add whatever pressure I can. So here is a letter I plan to send the FDA and maybe my Congressor once I figure out who that is. Please let me know about any factual errors, logical errors, or things I should change because they are unconvincing or offensive:

To whom it may concern:

I am writing to voice concern at your decision to ask the company 23andMe to halt genetic testing.

As a doctor, I am well aware of both the importance of genetic testing in medicine and of the difficulty in getting these tests through the normal medical system. Many of my colleagues do not know of the existence of personal genomics, and believe that genetic testing is a once-in-a-blue-moon procedure to be ordered only when there is high clinical suspicion for a rare disease such as cystic fibrosis.

Even in cases where doctors are willing to use genetic information to help in treatment, they have few good options within the standard medical system. Tests for a single mutation may cost several hundred dollars, in addition to the expensive consultation with a physician necessary to order the test and the second expensive consultation with a physician necessary to receive results from that same physician. These are rarely covered by insurance, especially for poorer patients. Patients are rarely willing to purchase these expensive tests out-of-pocket, and when they do, both patient and doctor must wait weeks for each individual result they want.

In contrast, 23andMe has raised awareness of genetics among the general population and given them questions and concerns, usually appropriate, which they can discuss with their doctor. Their doctor can then follow up on these concerns. Such followup may involve reassurance, confirmation with other genetic testing, confirmation through other diagnostic modalities, or referral to another professional such as a genetic counselor. In my experience personal genomic results do not unilaterally determine a course of treatment, but may influence an ambiguous clinical picture in one direction or the other, or be a useful factor when deciding between otherwise equipotent medications. Banning the entire field of personal genomics in one fell swoop would eliminate a useful diagnostic tool from everyone except a few very wealthy patients.

Let me give a personal example. My family has no history of any genetic disease, I (like many Americans) do not have a primary care physician, and I had no reason to go to a doctor to get an expensive genetic screening. When 23andMe began offering genetic tests through the mail for $99, the easy availability convinced me to try one out. From that test I discovered that I was, like one in every thirty Ashkenazi Jews, a carrier for hereditary sensory and autonomic neuropathy type III, a serious and often fatal neurological disease linked to a locus on chromosome 9 which can be detected by genetic screening with an accuracy of >99%.

This has given me information I need to control my future: in particular, it has made me more wary of having children with any other Ashkenazi Jew until they have been tested for the mutation as well. Without personal genetic testing services such as 23andMe, I would not have discovered this mutation; with the loss of these services, my future partner will have to convince her primary care doctor to order her a prescription version of the test. Such a doctor will almost certainly never have heard of this rare condition, may blow her off, and may charge a hefty fee for this service if they perform it at all.

In addition to these immediate consequences, I am distressed by the likely effects of this decision on genetic research. Many of the most promising medical advances in the pipeline are based on genetics, but one major bottleneck to genetic discovery is the absence of good genomes to work with. A typical genetic experiment involves using advanced software to tease out small correlations between genotypes and phenotypes. Such experiments require genotypic and phenotypic data from thousands of individuals, which is often hard to come by especially given legal restrictions. 23andMe has amassed what may be the largest database of such information anywhere in the world and is making it available to researchers (with appropriate privacy protection), an amazing public service for a for-profit company. Restricting their ability to provide this service will almost certainly delay life-saving genetic discoveries.

In contrast to these important services provided by 23andMe, your stated worries about the company are, with all due respect, somewhat bizarre. As I gather from your letter to the company, you have at least three fears.

First, you state that “if the BRCA-related risk assessment for breast or ovarian cancer reports a false positive, it could lead a patient to undergo prophylactic surgery, chemoprevention, intensive screening, or other morbidity-inducing actions”. As I would hope the FDA is aware, it is extremely common in medicine to screen for rare conditions through affordable and noninvasive tests, with positive results followed up by more expensive, more complicated tests to confirm the diagnosis. You allow, for example, prostate specific antigen testing, even though the positive predictive value of this test at its most commonly used values is only 30%, because you correctly assume it will be followed up by further analysis before any drastic steps are taken. If the FDA believes that there are surgeons who would say “You claim to have heard you had a BRCA mutation, I have no follow-up questions, let’s schedule you for a double mastectomy”, then the FDA should be pursuing those surgeons, not genomics companies.

Second, you state that “we still do not have any assurance that the firm has analytically or clinically validated the PGS for its intended uses, which have expanded from the uses that the firm identified in its submissions”. If this means only that you are uncertain whether 23andMe’s genetic testing actually reads the genome correctly – that what it reports as an A nucleotide is indeed an A nucleotide – then I agree this is a reasonable question for the FDA to pursue. However, by your mention of intended uses “expanding”, I gather that in fact you mean the specific genetic claims they make, for example that genetics affect warfarin metabolism.

There are several thousand papers about genetics, linking specific genes to everything from eye color to intelligence to chance of developing colon cancer. Some of these papers are no doubt correct, others will no doubt in time be found incorrect. However, I worry you want 23andMe to independently validate each of these results before informing customers about them. Worse still would be a demand that the company validate the results *with its specific testing kit*. It would be cost-prohibitive even for a single company like 23andMe. It would destroy any possibility of competition; even if one company did it, new startups would not be able to. And it is unnecessary. If BRCA is linked to breast cancer in general, and 23andMe’s kit reads the genome correctly, then 23andMe’s result for BRCA will be linked to breast cancer, as will the result of any other kit that reads off the genome correctly. Asking companies to validate each result for their particular kit makes no more sense than asking each pharmacy to do independent studies to confirm that their own bottle of penicillin kills bacteria. The pharmacy’s job is to prove they have real penicillin: once they do that, the general result that penicillin kills bacteria applies.

In my opinion, 23andMe handles the complexity and uncertainty of genetic testing entirely correctly. It rates each of its tests on a “confidence” scale showing how much consensus there is in the literature that the result is correct and gives a clear explanation of what each confidence level means, citing all the studies it uses to gather data. It states very clearly on every single page that its results are open to interpretation and should not be used to unilaterally guide medical treatment. On results with less than its highest level of confidence, it refuses to even give numbers out of a fear that these might seem unfairly authoritative, only linking to the appropriate study and allowing consumers or their doctors to draw their own conclusions.

Third and most important, you state that “assessments for drug responses carry the risks that patients relying on such tests may begin to self-manage their treatments through dose changes or even abandon certain therapies depending on the outcome of the assessment”. For me, this is the most worrying objection, because it seems to acknowledge that although 23andMe may provide accurate information with the appropriate disclaimers, merely giving patients access to medical information is itself dangerous.

I agree that this can sometimes be true. For example, when I was in medical school many of my classmates came down with so-called “medical student syndrome”, in which their increased awareness of symptoms and diseases made them more likely to believe they had some rare and dangerous syndrome, causing unnecessary pain and sometimes even doctors visits and lab tests.

But the alternative is banning all information that anyone could possibly use to evaluate their own health. I have even heard the tongue-in-cheek proposal of banning medical textbooks. You worry that if a patient learns their genetic susceptibility to warfarin it may convince them to alter their own warfarin dose without a doctor’s input. But the same is true if a patient learns the (true) fact that leafy green vegetables affect warfarin levels, or that women require less warfarin than men – facts that can be discovered in any book about medicine or pharmacology. Likewise, knowing one’s weight may convince someone to take unsafe diet pills, to starve themselves, or for that matter to adjust warfarin dosing – the clearance of anticoagulants appears to increase with weight. But banning people from weighing themselves without a prescription is neither clinically nor ethically sound.

The two consistent options are to work to ban patients from learning any accurate information that might be relevant to their health – exactly the opposite of ethical medical practice – or to trust patients and doctors to work together to use information responsibly to support their own care, whether that information comes from a bathroom scale or a genetic test.

In conclusion, while I am excited about the possibility of the FDA helping patients and doctors by vouching for the accuracy of the general process that personal genomics services use to read the genome, I am extremely concerned that it may go beyond this useful service to place unnecessary or impractical restrictions that shut down the field of personal genomics entirely and prevent individuals and medical professionals from getting inexpensive, simple, and unrestricted genetic testing. I urge you to continue negotiating with 23andMe and to take every step possible to ensure that their important service returns to the market as soon as possible.

Thank you for your time.

Yours,
Scott [Alexander], MD, BCh, BAO

Sort of wondering whether I should cite sources for claims like HSAN3 being detectable with genetic testing or warfarin metabolism changing with weight. Is it considered appropriate to have a references section in a letter to a congressional representative? Inappropriate enough to counterbalance the level of awesome?

EDIT: Looks like there’s a petition to sign.

EDIT2: Madeleine Ball points out the less immediately sexy but ultimately equally important issue of the NIH Genomic Data Sharing Policy, which would affect some of these same things but is getting a lot less press.