What Five FDA Letters Mean for the Future of DTC Genetic Testing
The FDA has published online letters sent to five personal genomics companies – 23andMe, Navigenics, deCODE Genetics, Knome and Illumina – informing the companies that they are manufacturing and selling medical devices without appropriate FDA premarket review and approval. No surprise that the news that the FDA has sent out letters to some of the most well-known providers of DTC genetic testing products is already making waves. (Daniel MacArthur was the first to point me to the AP story, and Mary Carmichael of Newsweek and Andrew Pollack of The New York Times were among the first to dive into the substance of the letters.)
Below, we will discuss the immediate and long-term implications of the FDA’s most recent regulatory actions for the five companies receiving letters, as well as for the DTC genetic testing industry. First, however, a review of the letters themselves is required. Each of the five two-page letters is signed by Alberto Gutierrez, Director of the FDA’s Office of In Vitro Diagnostic Device Evaluation and Safety (OIVD), and follows a similar format throughout. To gauge the impact of these letters we will take them paragraph by paragraph.
1. The Devices in Question. Each letter begins with a description of the product the FDA has determined qualifies as a device under Section 201(h) of the Federal Food, Drug and Cosmetic Act (FDCA). Section 201(h)(2) of the FDCA defines “device” as:
an instrument, apparatus, implement, machine, contrivance, implant, in vitro reagent, or other similar or related article, including any component, part, or accessory, which is…intended for use in the diagnosis of disease or other conditions, or in the cure, mitigation, treatment, or prevention of disease, in man or other animals
The agency’s letter focuses in on the second half of the 201(h)(2) definition, and adds in language from 201(h)(3) which allows the FDA to categorize products as a device if they are “intended to affect the structure or any function of the body of man.” The “devices” identified by the FDA are: 23andMe Personal Genome Service™, Navigenics Health Compass, deCODEme Complete Scan, KnomeComplete™ and the Illumina® Infinium HumanHapp 550 array. It’s difficult to see how any of these products would fall under 201(h)(3) but not 201(h)(2) but, regardless, the FDA is making a clear statement: based on the plain language of the FDCA, it considers these products to be medical devices.
2. Has Your Doctor Seen This? The second paragraph, which appears to be identical in every letter, is a generic overview of the Medical Device Amendments (MDA) to the FDCA. The MDA grants the FDA the authority to require premarket regulation of medical devices. The letter stresses the importance of demonstrating both analytical and clinical validity to ensure that “individuals are not misled by incorrect test results or unsupported clinical interpretations” and that the products are “used to support good healthcare decisions.”
We wrote earlier this week about the difference between analytical and clinical validity, and the role that CLIA (which is administered by CMS, not the FDA) is meant to play in ensuring the former. The letters clearly indicate that the FDA has its eye on both measurements of genetic test quality, as it should. More significantly, the letters appear to indicate that the agency considers the products in question to provide “clinical interpretations” and/or to be used in “healthcare decisions.” In the long-running debate over whether DTC genetic testing qualifies as clinical medicine, it appears that the FDA may have come down on the clinical side of the fence, at least for these particular products.
In fact, Director Gutierrez clarified this point in an interview this afternoon with Newsweek. Director Gutierrez seemed particularly concerned with products that report on genetic variants related to drug metabolization:
If you’re making a claim about [a genetic variant that affects the metabolism of the anticoagulant drug] warfarin, and somebody decides based on the result they get that they want to change their dosing, that is a fairly risky decision. That could affect their health. If they’re not feeling well on their current dose and the drug is expensive, we don’t know what they would do.
These concerns are reiterated, albeit in less detail, in several of the actual letters (see next paragraph).
3. Where’s the Approval? The third paragraph in four of the five letters informs the company that the FDA has not received any information pertaining to the analytical or clinical validity of the products for use in the FDA’s “clearance or approval” of the products. The letters go on to describe the type of genetic information provided by the companies that is of particular concern in this regard, for example the warfarin and clopidogrel response information reported by 23andMe and Navigenics and the breast cancer risk and detection information provided by deCODE. The FDA, echoing Gutierrez’s comments above, warns that “consumers may make medical decisions in reliance on this information.”
The use of the word “consumer” is an interesting choice, particularly in light of the focus that the FDA is clearly placing on the potential clinical use of the products. It is particularly confusing in certain letters, 23andMe’s for instance, where the recipient of the genetic information is referred to in the same paragraph as both a “patient” and as a “consumer.” Clearly, one task for the future regulation and description of genetic testing products will be to clean up the terminology.
It’s also important to point out that Illumina receives special treatment in this paragraph. Of all the companies receiving letters, Illumina is the only one not to offer at least one product directly to consumers (the company does offer a commercial whole-genome sequencing service, although that product requires the participation of a healthcare professional). Illumina’s letter notes that the company has “received FDA clearance or approval for several of its devices,” but not for the HumanHap550 array. “Yet Illumina is knowingly providing the HumanHap550 array to 23andMe and deCODE Genetics for clinical diagnostic use without FDA clearance or approval” despite its being labeled “For Research Use Only.” As Gutierrez notes, “…Illumina has to follow the law, and they are aware that the chips are not being used for research only.”
Again, it appears that the FDA’s determination that the products are being used for clinical diagnostic purposes and delivered directly to consumers appear to be important factors, with the FDA drawing no enforcement distinction between enabling DTC genetic testing and providing the DTC genetic tests themselves.
4. We Know What You Said Last Summer. All but one of the letters then go on to describe a meeting last summer between the company in question and the FDA (July 29th for 23andMe, July 31st for Illumina, August 6th for Knome and August 13th for deCODE.) Surprisingly, there is no mention of a Navigenics / FDA meeting although, in part due to this paragraph’s conspicuous absence from the Navigenics letter, I wonder if this may have been an inadvertent omission.
During these summer meetings the companies presented information about their products to the FDA. On the basis of that information, as well as other available information (e.g., the FDA notes that 23andMe has “recently begun distributing the collection kit for your device through a third party distributor, Amazon.com”), the FDA has concluded that the products in question are diagnostic devices subject to FDCA regulation.
In its letters to 23andMe, Knome and deCODE, the agency goes on to explain that, since the products are “not developed by and used in a single laboratory,” it does not consider them to qualify as laboratory developed tests (LDTs). This is an important point because the FDA has long exercised “enforcement discretion” over LDTs, choosing generally not to regulate this category of test, one which includes a majority of currently available genetic tests. Since so many genetic test providers – not just those of the DTC variety – have relied on the LDT determination as a basis for skirting FDA regulation, this section in particular is likely to raise the blood pressure for companies that purport to offer one or more LDTs but do not conduct all of their development, testing and interpretation in house.
5. Where is Your Letter? Next is a paragraph, nearly identical across all five letters, that reminds the companies they have not received what FDA believes to be the necessary regulatory approvals. Or, as the agency puts it, “we are not aware that you have an approved application for premarket approval (PMA) in effect pursuant to” the FDCA, nor have you “notified the agency of your intent to introduce the device into commercial distribution as required by section 510(k)” of the FDCA.
Briefly, the FDA regulates medical devices in three classes.
Class I devices are simple devices that pose a minimal risk, and are generally exempt from FDA premarket approval or clearance. However, registration of the device is required (as is true of all FDA regulated devices).
Class II devices represent an intermediate level of risk, and require regulatory clearance (as opposed to an “approval”) before they can be sold in commerce. The FDA must determine that the “device to be marketed is as safe and effective, that is, substantially equivalent (SE), to a legally marketed device not subject to premarket approval.” The clearance track for Class II devices is set forth in section 510(k) of the FDCA.
Class III devices are the riskiest device class, and the products that receive the most stringent FDA scrutiny. FDA approval is required before the device can be sold in commerce, and is granted only when the FDA determines that there is “sufficient valid scientific evidence…that the device is safe and effective for its intended use.” The regulatory submission is substantially more detailed than under 510(k), and the FDA is not obligated to respond as quickly.
The FDA letters provide no real insight into whether the agency considers the products it has identified to represent Class I, Class II or Class III devices, a classification that will determine the size of the regulatory burden imposed by the FDA.
6. We Will Be Waiting. Finally, the FDA advises each company to “take prompt action to respond to this letter” and offers to meet with the company to “discuss whether there are tests you are promoting that do not require review by FDA…”
“Prompt” is not defined, and most or all of these companies have had open channels of communication with the FDA for some time now, so it is unclear what type of timetable this imposes. It is clear, however, that this is an invitation to further agency dialogue, and that these companies decline only at their peril.
The second part of the paragraph – the agency’s offer to “discuss whether there are tests you are promoting that do not require review by FDA” – strikes us as a possible opening, at least for some of the identified companies and tests. Last fall, 23andMe, following in the footsteps of its competitor (and recent FDA regulatory target) Pathway Genomics, announced that it was breaking up its genetic testing service so that it could offer separate products for customers seeking to explore their genetic ancestry but who were not interested in the more medical applications of personal genetics, or vice versa. (23andMe also offers a combined product that includes all of those features). As I wrote at the time, that is just the sort of distinction that might be significant to the FDA as the companies and the agency discuss which specific products require premarket clearance and approval (more on this below).
What Does It Mean For the Five Named Companies? The immediate implications of the FDA’s letters may be less significant than some might initially suspect. After years of speculation about whether and how the FDA would regulate DTC genetic testing products, the agency has now publicly delivered at least a partial answer: it considers these specific products to be medical devices requiring either premarket clearance or approval, and it does not consider them to be LDTs subject to regulatory enforcement discretion.
For the companies named in the letters, at least, this provides a concrete agency determination to which they can react. It’s unlikely that the response from any of the companies will be to pull their products completely off of the market and, as The New York Times reports, Director Gutierrez has indicated that “it would be unfair to remove the tests from the market because the agency had not, until now, clearly told the companies that the devices needed approval.”
[Added in Edit, 6/11: Turna Ray of Pharmacogenomics Reporter has published her own recap, which contains additional comments from Erica Jefferson, an FDA press officer. Jefferson's comments make a point of distinguishing the five untitled "letters to industry" from "warning letters." Jefferson explained the difference between the two types of letters:
While Warning Letters set out specific violations of law that a company must address immediately or else the agency will take an enforcement action, an Untitled Letter identifies agency concerns and gives a company the opportunity to meet with the agency and to have time to take appropriate steps to address these concerns...Based on how the companies respond to the Untitled Letters, FDA may follow up by sending Warning Letters.
Jefferson also added that the letters were sent based in part on recent discussions between the FDA and several of the companies that took place in the aftermath of the FDA's decision to send Pathway Genomics a similar letter last month. According to Jefferson, those meetings "helped inform the agency's decision to send the Untitled Letters." End Edit, 6/11]
So, at least for the moment, we may see little or no immediate change while these companies weigh their options internally and through discussions with the FDA. What exactly are those options? They obviously vary based on the specific company and product, but here are a few of the most likely possibilities:
Wave Goodbye. For products that have failed to meet expectations, or are no longer an integral part of the company’s future plans, one possibility is to simply pull the product from the market. The most likely candidate for this response would appear to be the deCODEme test, which is considerably more expensive and less popular than a number of its competitors. deCODE Genetics recently emerged from a well-publicized bankruptcy, and there have been hints that deCODEme might not be part of the reorganized company’s long-term plans. (However, in comments today to The New York Times, neither deCODE or its head of research, Kari Stefansson, indicated that they would do anything other than cooperate with the FDA). If deCODE or any other company does decide to pull its test from the market, existing customers will likely be anxious to know what will happen with their genetic data.
Say Hello (to the FDA). For other companies (e.g., Illumina, a company with considerable experience navigating the FDA approval process), the path of least resistance may be to simply agree with the FDA and seek the appropriate clearance or approval. The viability of this option will depend on how the FDA intends to categorize the specific product (e.g., Class I, II or III) and whether the company believes (a) it can bear the burden imposed by such a regulatory submission and (b) that its product will be approved without changes to its substance or commercial availability that would materially undermine the product’s commercial viability.
Change Pathways. Perhaps the most palatable option for many of these companies is to consider altering the product in a manner that would convince the FDA it no longer qualifies as a device requiring premarket approval or clearance, for instance by removing the ability of consumers to purchase the product without the participation of a healthcare provider.
In his interview with Newsweek, Director Gutierrez discloses that Pathway Genomics was not sent a letter yesterday because the company responded to the agency’s previous letter and has indicated that “they are planning to move away from direct-to-consumer testing…” While that disclosure is news to me, and not one I believe the company has made publicly (the website still appears to allow consumers to purchase directly), it is not a surprising one. As Gutierrez notes, another genetic testing company, Counsyl, made a similar decision in the aftermath of the Pathway / Walgreens commotion.
Of the five companies that received FDA letters this week, Navigenics and Knome would appear to be the most likely candidates to pursue this option. Navigenics has increasingly shifted its product focus over the past year in the direction of more traditional medical channels (as evidenced by its receipt earlier this year of a clinical laboratory permit from the State of New York, apparently the first awarded to a personal genomics provider), and this development could be the final nudge it needs to pull the plug on its DTC option. Similarly, Knome, which once offered whole-genome sequencing directly to the super-rich for $350,000, has increasingly positioned itself as a provider of genomic software and interpretation, with a focus on research and clinical applications. As with Navigenics, if Knome can forestall FDA regulation by eliminating all DTC versions of its products, it may have a strong incentive to do so.
A related approach, discussed above, would be for companies to seek different regulatory treatment for products that have clearly different uses. 23andMe, for instance, might seek a different classification for its ancestry testing product as compared to its products that provide genetic testing of a more arguably medical variety, such as disease prediction or drug response.
Prepare for War. Finally, there’s always the possibility that one or more of the companies will challenge the FDA’s determination that they are either (a) offering a medical device or (b) not offering an LDT. There’s no question that going toe-to-toe with the FDA represents the path of greatest resistance, but if any of these companies feel sufficiently backed into a corner by the FDA’s approach this could surface as a viable option.
While no company has yet indicated its intent to challenge the FDA’s interpretation, 23andMe has thus far been the most outspoken in its criticism of the agency’s recent actions. As reported by The New York Times, one 23andMe director suggested that denying consumers direct access to their genetic information would be “appallingly paternalistic” (a characterization Director Gutierrez found inapplicable to the FDA’s regulatory decision), and the company has indicated that it “disagree[s] with the FDA’s conclusion” but is “open to discussion on ways to regulate the personal genetics industry.” Only time will tell whether 23andMe, or some other party, attempts to challenge the FDA’s regulatory approach to DTC genetic testing.
What Does it Mean for the Rest of the DTC Genetic Testing Industry? For the rest of the industry, the regulatory outlook is little clearer today than it was yesterday. The FDA has offered specific regulatory determinations for a limited set of DTC genetic testing products, but it has not offered broader industry guidance.
From the letters, and from Director Gutierrez’s statements, it is clear enough that the agency considered several important factors in identifying these five specific companies and products as regulatory targets. These include the DTC availability of the product (or, in the case of Illumina, contribution to DTC availability), the perceived medical use of the product and, in all likelihood, the complexity of the testing and interpretation involved in the product.
But how the FDA weighed those factors against others – including the utility of the tests, the reality of its limited regulatory resources, and the presence of numerous other genetic tests offered to consumers and to patients – remains unclear. Keep in mind that the FDA sent letters to five companies that, while they represent some of the best known genetic testing providers, do not comprise the entire DTC genetic testing industry (see, for example, this list at DNA Test Index or this list at AccessDNA). For other DTC companies, as well as companies and investors seeking to break into the DTC marketplace, there continues to be a lack of clarity into the FDA’s DTC genetic testing regulatory strategy.
For that reason and others, my own opinion continues to be that transparency – and not regulation – is what would be most beneficial to the DTC genetic testing industry and its customers at this time. Until companies, consumers and regulators better understand the tests that are available and, importantly, how those tests are being used, it will be difficult to develop a regulatory policy that protects the health and safety of individuals without stifling commercial innovation and individual exploration. In the meantime, expect the FDA’s latest actions – as well as, possibly, the ongoing Congressional investigation – to significantly shake up the personal genomics landscape in the coming weeks and months.
Dan,
Nice post. However, what I have been saying since Feb 2009 about BRCA DTCG testing is clear. Medicine is medicine, not matter how you spin it. The companies should develop strategies to join the community rather than shun it. Medicine requires regulation,
and it would be foolhardy to fight that stance, no matter how many Googlebackers you have. Time to be responsible here.
Dan,
Further, if you look at FDA approval of Amplichip, they called it a class II device. I am certain the same will be true of these. http://www.fda.gov/MedicalDevices/ProductsandMedicalProcedures/DeviceApprovalsandClearances/Recently-ApprovedDevices/ucm078879.htm
Further, by legal statute: All devices in existence after 1976, once classified as a medical device are placed in calss 3 category UNTIL reviewed by the FDA and reclassified.
I am not so certain they can be out there selling these tests without reclassification. You may want to double check, but that is what Roche’s approval letter and legal statue says.
FYI
-Steve
Dan,
The most drastic effect I can envision as a consequence of this move by the FDA, is a significant disempowering of the Patient Empowerment movement. Since I consider the latter to be precious aspect of proper 21st century healthcare, I believe all will suffer by blind acceptance of this ruling. I’m afraid I favor the last option on the list of the options you provided- Prepare for War. Not a bloody one, but a highly-sophisticated response that might include, for example, some subtler forms of compromise in terms of the presentation of the data to the patient-consumer.
While 23andMe has done some dodging about on this overall regulatory issue, I agree with their basic premise that the consumer has the right to their own DNA data, without FDA intervention, even if some of the test results happen to have clinically-related implications. I was pleased to see the initial response by 23andMe to the FDA was not fully conciliatory. My comments are with all due respect to the FDA, who are just trying to do the proper job.
Bob
Dan – thank you for the excellent information. There are so many questions, here are a few that come to mind:
1. They want to assess analytical and clinical validity – OK, but does it just refer to DTC? What about tests developed and sold through MDs such as the offering via Steve Murphy’s Genetichub or the recent Existence Genetics (especially the latter which uses an undefined proprietary chip plus various software and algorithms they have developed). How are these regulated?
2. FDA said software is a device. Where does that leave 3rd party analysis tools such as Promethease?
3. Any effect on the Personal Genome Project – or is that defined as research only?
4. If a company gets clearance but then updates their service with new SNPs or new interpretations (and the latter happens constantly), what then? New clearance required?
5. If decodeme sold their service from Iceland with marketing purely via internet on non-USA servers could they continue without the need to engage the FDA?
Loads more, but I don’t want to burden you with them all and I think you are going to be busy!
Gutierrez : “If you’re making a claim about [a genetic variant that affects the metabolism of the anticoagulant drug] warfarin, and somebody decides based on the result they get that they want to change their dosing, that is a fairly risky decision. That could affect their health. If they’re not feeling well on their current dose and the drug is expensive, we don’t know what they would do.”
Somebody has paid for a genetic test presumably having read what it’s all about, somebody has received results and read what they are all about (must have if has discovered potential effect on warfarin). Somebody, having got this far) is intelligent enough to know that changing medicine dosage on one’s own is a risky thing to do (not rocket science). Just like many other aspects of life.
What if somebody is feeling unwell on warfarin, reads about sensitivity of dosage on WebMD and decides to autotitrate (without genetic information)??
Also – there are numerous medical sites with risk calculators and disease diagnosis algorithms, these are software, will they also be classed as medical devices? Potentially lead to risky decisions, no?
Keith,
The way I read this law is pretty wide open, with lots of discretion. The FDA has had the power to do these things for quite some time. Dan may be right, this may be an Alamo for one of these companies. Namely, the company who has built an empire on information.
PGP is research and not considered part of this.
GH is test agnostic and uses only Clinically Validated labs.
When a test changes, just review the FDA precedent, it is a simpler process.
Your Iceland question is similar to a shell game. The consumer is considered the location of transaction. Thus, just like drugs made in other countries, FDA will have some say. Maybe not ability to enforce, but ability to have a say.
Promethease? Heck, FDA is looking to make EMRs into medical devices. In some instances they are correct. In others not so much. But, just look at all the iPhone apps out there that meet device criteria. It is a HUGE space!
As I said, DTCG will be classified as a class 2 device and until review is legally a class III.
Dan, would love your thoughts here……
Keith and Steve: I think the majority of your questions are, unfortunately, unanswerable at this time. Steve is correct in that the FDA’s discretion appears to be fairly broad, at least until it is more narrowly defined by either Congress or the courts. That discretion appears to extend to, among other topics, which LDTs will require approval and which DTC tests will be classified as devices.
For what it’s worth, I am hearing that the agency would likely treat at least some of the products it identified in the above letters as Class II devices, as Steve has suggested. Of course, that presupposes that one or more of the companies proceeds with a regulatory submission, which is not a given. Still, until we have a better understanding of all the factors the FDA is considering, and how they weigh those factors against each other, I’m not sure we have much more than case-by-case evaluations (including in a number of the examples Keith raises, particularly in the software and research contexts).
The software discussion will be a particularly important one, since as the production of genetic data becomes commoditized the product differentiation will be at the level of information interpretation. How the FDA will draw the line between device (e.g., Knome) and informational resource (Promethease, WebMD, Google, the growing number of medical iPhone apps Steve mentions, etc.) is unclear.
In my opinion, all of this argues in favor of a standardized, mandatory registry at this point in time.
Finally, Bob, no matter what this means for personal genomics in the short term, I think that over the long term (5+ years) this is likely to be a blip on the radar screen when it comes to individuals having unfettered access to all of their information, not just genetic or even medical data.
Thanks for the useful analysis, Daniel. I disagree with your bottom line (I support some regulation), but your discussion was clear and useful.
I am drawn toward Steven Murphy’s conclusion that these are Class 2 devices, which require “special controls” in addition to Class 1′s “general controls.” I don’t know whether the FDA has ever tried this as a “special control,” but one might think about a requirement that a qualified health care provider (physician, clinical geneticist, genetic counselor) in a health care provider/patient relationship with the consumer (a fiduciary for the consumer and not just the company’s employee or independent contractor) be necessarily involved in the interpretation and transmission of test results. That, it seems to me, could be the crucial measure needed to help ensure that these results are used safely – a perfect kind of Class 2 special control, just like special precautions with motorized wheelchairs.
I wonder whether 23andMe will be foolish enough to go to war with the FDA. They might be – they seem to believe their libertarian take on “my own genome.” I think 23andMe, and its supporters, risk becoming a left or libertarian version of the 1970s and 1980s fans of treating cancer with Laetrile and coffee enemas. I suggest that is NOT a good place to be, for the advocates or for the patients.
Hank Greely
Professor Greely, can you clarify your thinking on why I need approval from some third party to look at my own body and my own genes?
I would put these reports somewhere between oral thermometers and pregnancy tests wrt how much regulation is necessary. Requiring a doctor to interpret everything is akin to requiring a auto mechanic to read my OBDII scanner -unnecesary,obtrusive, and un-American.
@Jeremy,
Oral Thermometers and Pregnancy Kits are not the same as these tests at all.
Oral thermometers give you biometric data. They may even tell you your temperature is elevated. But they do not tell you what medications will or will not work to lower your temperature. Further, they too have some form of regulation and hurdles to overcome before being sold.
Pregnancy kits are also regulated by the FDA.
Lastly, Microarrays that provide some diagnosis are judged to be Class II medical devices. This was decided in 2005.
The FDA is following laws and regulating according to those laws.
We do live in an America that is governed by the rule of law. If you have a problem with the laws, you can always vote.
But I am certain these devices ARE medical devices and will be regulated as Class II devices.
Dan,
Thanks for your analysis – very helpful.
I’m wondering if you’d be willing to briefly discuss how you think this development impacts potential regulatory changes for LDTs? You mention that FDA has “long exercised “enforcement discretion” over LDTs, choosing generally not to regulate this category of test.”
However in this decision, FDA is calling out “clinical validity” and the potential for test results to impact clinical decisions. This, obviously, applies to every CLIA-lab LDT as much or more than these DTCG tests.
Do you think that these letters increase the chance that FDA may change its stance on LDTs in the medium term?
Thanks,
Gavin (Genentech)
OK so the FDA is worried about people taking medical decisions on their own. Misha Angrist did a nice post on this at Genomeboy (http://bit.ly/a5iLZ9). But it’s not just medication according to the NYT: “It is not unknown for women to take out their ovaries if they are at high risk of ovarian cancer,” Dr. Gutierrez said. What? On their own? Their is a danger that they get their 23andme results then go into the kitchen and whip out their ovaries??? I hope that mine doesn’t show a higher risk of prostate cancer…
Oh come on, there may be some valid reasons to want to exert some control but this is insane.
As Misha, myself, and many others have pointed out, there are many, many other sites, even official govt sites, that give personal medical advice – on warfarin dosing (which even includes genetics), breast cancer risk (dammit, back to the kitchen), heart attack risk, and so on…
The point is not whether they need to be regulated or not, it’s about what is feasible, effective and useful. Software is a Class II device? It seems that the FDA have tried to account for the future where genotyping will be routine and interpretation services will not require a lab test. But that is no different from any of the current medical advice sites that use algorithms that calculate risk incorporating biological information from an individual.
There are better alternatives as proposed by Dan in a previous post – transparency, genetic register, there is a good code of practice piloted by the UK’s HGC (see http://bit.ly/aIahOR). There is a problem though, the CoP is supported by all the main commercial players but so far, 18 months later, no progress has been made. The FDA would do well to take a step back and use their stick to demand a) a strong CoP, b) Genetic testing registry c) setting up of some sort of peer review type of editorial board to provide independent assessments of the various tests on offer and scrutinise them for accuracy.
Openness and scrutiny has worked so far – there is no evidence of any harm. Of course the FDA and the rest can some up with an endless list of “could happens” but guess what, most people are not stupid. DTC testing, including for Pgx, has been around almost a decade with little evidence of self harm or self medication because of it.
Sorry one more: But Dr. Gutierrez denied that the agency was being paternalistic. “We really don’t have any issues with denying people information,” he said. “We just want to make sure the information they are given is correct.”
The current situation (at least for the sites like 23andme, decode & navigenics) more or less guarantees that.
Terrific discussion!
I have nothing but respect for Hank Greely, but I think his comment comparing 23andMe’s service to Laetrile points up the lose-lose-lose for DTC genomics companies. If they return information on earwax or ancestry, then they are frivolously catering to well-heeled narcissists. If they return information on type 2 diabetes or other conditions where the odds ratios don’t move much, then they fail the clinical validity test and are charlatans, even if the information is accurate insofar as we understand it today. But if they return indisputably clinically valid information on BRCA or warfarin dosing, then they are practicing medicine without a license and must either involve physicians or cease and desist.
If one accepts these arguments, then it seems to me one is accepting the status quo: genetic information, if it is of any consequence (as decided by the medical profession), should only travel via the traditional, top-down path: patient sees doctor, doctor orders test, doctor interprets test, doctor returns result. Never mind that I know more about the test than my doctor, that I don’t want to wait 6 months or a year or drive 100 miles to see a geneticist, that even my geneticist will likely “diagnose and adios,” that I have my own reasons for wanting to know my APOE genotype, etc. Whwther a genome is $5000 or $50, this system is unsustainable.
why should the rules be different for ancestry applications when they would provide diagnostic information anyway. if whole-genome scans are used the infomation has been gathered it’s just a matter of interpretation. where do you draw the line?
Ditto Misha’s point about Hank Greely’s comment. I tweeted Dr. Greely’s excellent Viewpoint in last month’s The Lancet (“Challenges in the clinical application of whole-genome
sequencing”; http://tinyurl.com/36mnz44), which I felt was timely material for (particularly) med students in training, but I do not agree with his comments in this post. Also, still hoping he responds to the pointed question by “Jane Murphy”. Respectfully,…
Thermometers are FDA “exempt” Class I products. Pregnancy kits are Class II devices. Neither requires a doctor meddling to obtain one and access the results, which is really the issue here.
The FDA needs to lay out some loose guidelines and then get out of the way. If they try to get involved there will be a reversion to a hobbyist or hacker culture in which companies sell you your raw sequence and then it is up to you to submit it to various websites located in Eastern Europe to get any kind of interpretation. This is not the direction we want to head in.
@Jeremy,
DTCG will be class II and subject to premarket review. I am not certain they will ask for MDs consult, but they likely will put these tests off the market for a year or so. If the companies can last that, they will survive.
And we will all be assured of a safe and reliable product. Is that such a big deal?
Oh, I get it, break a few eggs to make an omelette…..
But that is not how the FDA rolls……
Maybe 2 years for premarket, but that will be a def. especially if they won’t involve doctors.
A pregnancy test is not a complicated multifactorial algorithm to determine risk either. So stop with the pregnancy test and glucometer arguments, please. It is either ignorant or disingenuous.
-Steve
@Hank Greely: To clarify, I do not think that DTC genetic tests – or genetic tests of any kind – should be exempt from all regulation. These tests are already subject to CLIA and, as I wrote last week, it may be appropriate to expand the CLIA regulations with respect to genetic tests.
It may also be the case that certain genetic tests present sufficient risk to purchasers that they should be subject to additional controls, including the types of controls available to the FDA by declaring a test a medical device (e.g., Class II or even Class III designations).
However, I do not believe that the mere fact that a test is provided directly to a consumer is the proper criteria for determining whether a test requires additional regulation. To do so, at least under our current regulatory scheme, where the vast majority of LDTs are not FDA-regulated, implies that these tests are categorically safe when ordered by and delivered to a doctor, and categorically dangerous when there is no physician intermediary. I do not think the evidence at hand supports that distinction.
Instead, I think the appropriate step at this point is to gather much more specific data about the tests on the market, focusing on how such tests are actually used (or not) by both healthcare professionals and individuals to make medical decisions. If the FDA is concerned that “consumers may make medical decisions in reliance on” the information contained in these tests, shouldn’t we take steps to determine if this is in fact happening, rather than merely speculate?
I do not believe we have that data yet and, largely for that reason, I think the FDA’s current regulatory proposal may be premature.
@Steve Murphy: If the FDA does manage to regulate these tests, I agree that the most likely designation for many current tests will be Class II. However, I disagree that the FDA “likely will put these tests off the market for a year or so.” In its second attempt at IVDMIA regulatory guidance (see here) the FDA offered a 12 month grace period for all “currently marketed laboratory-developed IVDMIAs,” with an additional 6 month grace period for manufacturers submitting a 510(k) or PMA within the initial 12 months. I would expect to see something similar should the FDA proceed with broad-based regulation of DTC genetic tests.
@Gavin Williams: This latest development does not necessarily mean that the FDA is moving to revive prior regulatory efforts aimed at LDTs (particularly IVDMIAs), but I agree that the focus on clinical validity may make it somewhat more difficult to support direct regulation of DTC genetic tests alongside the continued exercise of enforcement discretion with respect to LDTs that are clearly designed for clinical use. I will go into the IVDMIA / DTC issue in much more detail in a post tomorrow morning, so please keep an eye out for that.
Thanks all.
- Dan
Past experience suggests that working and cooperating with the FDA will not lead to tests being removed from the market – some modifications to the claims and interpretations may be required to avoid being classified as a class II in the short term following which companies may, if they wish, apply for premarket review for the more “medical” interpretations. Even then they can still be DTC (which is important – there is also the issue of privacy, many do not want anybody else, including docs, to have their genetic test results and many certainly don’t want to have them stored in their medical records).
Given the complexity of these services, as alluded to by Steve, any premarket review will be a lengthy process, removal of the services until completion of review would kill the companies
@Dan – fully agree with your comments on LDTs, there is no logic. If a LDT is delivered through an MD it is deemed safe and uncontroversial – trust me, I’m a doctor??
Not sure but hadn’t Navigenics already stopped taking direct orders? I was on their website a few weeks ago and it didn’t appear that you could order directly, only through your doctor/job.
Sorry for the very long post, but there is lots here to say (especially with a relatively liberatarian crowd of commenters).
First, Jane Murphy’s question (“Professor Greely, can you clarify your thinking on why I need approval from some third party to look at my own body and my own genes?”) My answer is because we as a society do (and, I believe, should) protect people from certain risks, whether it is by requiring motorcycle helmets, registration of publicly traded securities with the SEC, or FDA approval for drugs. Medical information can be dangerous if misunderstood. I don’t know anything about you, in particular, and it’s certainly possible that you – and all the other commenters in this blog – would not misunderstand the information. But we don’t make regulations to protect the smartest and best educated among us, but the whole population. If there were a good way – logistically, legally, and politically – to exempt the well informed (“masters of the genetic universe”), then I’d have no objection to such exemptions. I like George Church’s requirement that people in the PGP pass his test with a score of 100% correct. (Misha, I assume you passed!) And, lastly, I would note that if I want to “look at my own body” in terms of getting a liver biopsy or a PSA test, I have to go through trained professionals who can minimize the risks, from the procedures but also from misunderstood information.
Second, to Misha. yes, I do think DTC companies are in a lose-lose-lose situation, though I wouldn’t, myself, use the most excellent and colorful words you used. (Well, maybe not.) I should note, though, that I’ve got no complaints (beyond the common false, or, at least, incomplete advertising) about the ancestry services (See Henry T. Greely, Genetic Genealogy: Genetics Meets the Marketplace, in REVISITING RACE IN A GENOMIC AGE 271-299 (eds. Barbara A. Koenig, Sandra Soo-Jin Lee, and Sarah Richardson; Rutgers University Press, 2008). If someone wants to pay for a genetic test for earwax type, fine. I think people should be able to pay for fancy sports cars, $2000 bottles of wine, or tickets to the NBA finals if they want.
But if the companies are using a business model that gives consumers medical information that has a reasonable chance of causing harm through misunderstandings, then they should lose.
This just isn’t a service, in my view, that should be provided DTC. It reminds me of the fad a few years ago for whole body CT scanning, which largely died out after some false advertising suits, negligence actions, and bankruptcies. If you are giving people powerful medical information, it SHOULD come with protections that are outside the normal DTC process, which currently means through clinical geneticists, genetic counselors, or doctors. (I do think, as I said in that Lancet piece, that we’ll need a new model for whole genome sequence information, which will overwhelm our current professional methods of providing good information, but that model needs to be invented.)
Part of what you are saying, Misha, is a complaint against current American medicine, one I agree with strongly. We should have a better medical system. (Frankly, as a very happy member of Kaiser Permanente Northern California, I think I do have a better medical system.) And part of what you say is that you don’t need the same protection as a lot of my relatives (and, I’ll bet, your relatives) do.But between making you either waste time and money on a meaningless professional (or making you find a good professional), on the one hand, and putting anybody who wanders into a Walmart (or the 23andMe on-line site) at risk, I’m going to put you through the hassles, every time.
I’m not worried about protecting you (after all, you passed George’s test!); give me a good way, as discussed above, of exempting you, and I’m fine with it. But a regulatory scheme isn’t just about you, or other well-informed people – it’s about over 300 million Americans, fewer than 1:10,000 (at best) of whom have your knowledge.
Finally, Dan, I didn’t say you didn’t believe in any regulation, though I can see how my statement that I disagree with your bottom line because “I believe in some regulation” could be read that way. Make that say “some FDA regulation.” I suspect we both support stronger CLIA regulation, though, unless it was much stronger, it wouldn’t eliminate, for me, the need for FDA regulation.
The rest of your argument, it seems to me, comes down to who has the burden of proof. You want the companies to be able to continue providing DTC results to consumers until the FDA proves it is dangerous. That’s the dietary supplement strategy of (non) regulation. I want the companies to have to prove that it is safe before they do it. That’s the general standard for drugs, biologics, and medical devices, in the US and elsewhere. The companies don the mantle of science, but they want to be treated like they are selling bee pollen or ephedra?
To be a bit more fair, I do think the LDT issue is hard, both for genetic tests and in general. Part of me would like all of them to be proven to be safe and effective (having clinical utility), through rigorous clinical trials, but I don’t think that’s practicable – especially with some of the rare genetic diseases – or politically feasible at all. So, in lieu of regulation that assures us that these tests (genetic and otherwise) really are safe and effective, I’m willing to go with a procedural intermediate – a requirement that someone talk with a trained professional about getting the test and about the test results, in the (sometimes futile) hope that the professional will help the (then consumer, now patient) avoid the more meaningless tests as well as avoid potentially dangerous misunderstandings of the test results.
I am not, in general, a fan of “the precautionary principle” – except for much of health care. If 23andMe wants to be a genetic genealogy company, I’m ok with that. If they want to get involved in health, then they should play by the rules we demand for most (although not all) of the health world.
Because we as a society do (and, I believe, should) protect people from certain risks, whether it is by requiring motorcycle helmets, registration of publicly traded securities with the SEC, or screening of personal genomic tests by trained professionals. Computers can be dangerous if misunderstood, or used improperly. There have been cases of suicide because of online bullying, and there is a massive amount of dangerous information on the internet such as pornographic images, incorrect medical advice, and hate speech. I don’t know anything about you, in particular, and it’s certainly possible that the well educated lawyers and health professionals supporting tight regulations of personal genomics know enough about computers and the internet to avoid these harmful situations. But we don’t make regulations to protect the smartest and best educated among us, but the whole population. If there were a good way – logistically, legally, and politically – to exempt the well informed (“masters of the computing universe”), then I’d have no objection to such exemptions. Possibly people should have to take a test before being able to use computers and the internet. If not, they should go through trained professionals, such as myself, who can minimize the risks from misunderstood or harmful information. I only charge $100 per hour.
Steve and Hank:
Do you have any legal argument to disagree with my analysis?
The FDA / USA government cannot restrict DTC Genetic Testing Companies from
a) analyzing our genome and providing us with our raw data, &
b) providing us with links to published scientific studies, which were often funded by the USA government itself via the NIH (National Institutes of Health), reporting “base … at SNP … is associated with a … higher/lower risk for disease …”.
Both a) and b) only provide factual information, and the FDA / USA government has no right to restrict the free flow of information (freedom of speech, which is protected by the First Amendment to the United States Constitution).
DTC Genetic Testing Companies also have the right to repost these published results in summarizing paragraphs, since access to the literature is controlled by high charges of the publishing houses.
Dirk, the FDA has statutory power over devices. One of the definitions of “device” is a (long list of nouns)
“intended for use in the diagnosis of disease or other conditions, or in the cure, mitigation, treatment, or prevention of disease, in human or other animals . . . . ”
If the FDA concludes a genetic test is “intended for use in the diagnosis of disease or other conditions,” then I think it has statutory jurisdiction. We don’t get to the question of the constitutionality of providing the information about health implications because it is the test that is disallowed – the proposed speech is irrelevant (except for use as evidence of the uses the test was intended for). That’s why I think that a forensic test, using the CODIS markers, or a genealogical test using mtDNA or Y chromosome STRs is unlikely to be a “device” subject to FDA regulation – they aren’t “intended for use to in the diagnosis of disease or other conditions, or in the cure, mitigation . . . [etc.]” You could sell Laetrile for use as a pigment for paint or as plant food, just if “intended for use in [a health context].”
Assuming there is statutory jurisdiction for the FDA, would it violate the commercial free speech doctrine to restrict a test that produces results that could then be the subjects of speech? That seems to me a stretch – I can’t think of any good parallels, but then, I’m not a First Amendment expert.
Ultimately, it seems to me that you have to fall back on a due process liberty right to control/know about your own body, which has been a winner in some contexts, like reproduction and death, but not in others, like drugs. That seems to me a “non-blushable” argument – one that a lawyer could make with a straight face – but one that would not be likely to win with this or any likely soon to come Supreme Court. But it is always hard to predict things, particularly the future (Niels Bohr) and what do I know – I picked the US to beat Ghana.
Hank:
So, the FDA has jurisdiction over ‘devices intended to diagnose a disease.’
One could argue that one has to distinguish between “genotype” and “phenotype.”
I think we can agree that a disease is a phenotype.
A person’s phenotype is mostly* determined by the structures and expression levels of his/her mRNAs and their encoded proteins, and the glycosylation of these proteins. Thus, one could argue that the FDA only has jurisdiction over devices that directly diagnose/investigate/probe a person’s mRNAs and proteins.
Maybe I should have become a lawyer.
*In addition to mRNA, other kinds of RNA also play a role: http://en.wikipedia.org/wiki/List_of_RNAs
Yes, genotypes are not diseases. Even when the genotype means the person has a disease, rather than having a higher risk for a disease, genotypes are not diseases, although one could still be using the genotype to diagnose the disease, in either case. (An elevated temperature is not a disease, but you use a thermometer to get the temperature to help diagnose a disease.) Allen Roses used to argue that APOE4 testing should be used, not to predict Alzheimer risk, but to help make a diagnosis of AD versus other dementias. (He probably still does.)
But let’s say we’re talking about a genotype that confers a risk that has not yet happened. When the company sells you your genetic information with genotypes that indicate such risks, is that information “intended for use in the diagnosis of disease or other conditions, or in the cure, mitigation, treatment, or prevention of disease”? Seems to me it probably is – especially if you argue that the information can help you prevent or mitigate the eventual disease. The hardest case might be something like Huntington disease testing, offered by someone and to someone who doesn’t have any hopes for prevention, mitigation, treatment, or cure. Even then, one might consider having, say, 60 CAG repeats in your huntingtin gene having a “condition” one might call pre-Huntington. (The “condition” language is for things that aren’t diseases but have medical implications, like pregnancy.)