As recently discussed on the DNA Exchange, general population screening for any health condition has both risks and benefits, and sound health policy should weigh all of these factors to ensure that we are doing more good than harm. But the debate about what, who, and how we should be screening for hereditary cancer risk seems to be mainly an academic exercise practiced by some in the clinical genetics and public health communities. While we continue to debate the merits of population screening, the commercial laboratories keep pushing forward, making these tests readily available to the general population through direct to consumer (DTC) testing products.
Often the cancer genetic test isn’t the main product that the customer is seeking, but rather may be a bonus of sorts provided along with ancestry and other DTC tests. Such is the case for the new health reports for MUTYH variants that 23andMe began releasing this past summer. Individuals who inherit one MUTYH pathogenic variant may have a slightly increased risk for colon cancer. Individuals who inherit biallelic pathogenic variants have a very high risk for colon polyps, colon cancer and other cancers and are said to have MUTYH-Associated Polyposis syndrome, or MAP.
Earlier this year, 23andMe received clearance from the FDA to release information about two variants in the MUTYH gene that are common in individuals of European ancestry – G396D and Y179C. The basis for this 510(k) approval is that it is substantively equivalent to the 23andMe BRCA test (which evaluates for three founder mutations most common in individuals of Ashkenazi Jewish ancestry). The company recently reported that they have over 10 million customers, with approximately half of those accessing the health portion of the testing (in contrast to ancestry alone). And considering that about 80% of those customers are of European ancestry, I estimate that of the approximately 4 million people received the new new MUTYH Genetic Health Risk report. Through this report. I estimate that about 250 people may have learned that they have MAP and about 60,000 customers learned that they are carriers for one of these two MUTYH variants.
So who should be screened for MUTYH variants? If you ask 23andMe the answer would be everyone. How should we screen? According to 23andMe, screening should be done by a two variant genotype panel. And who decides who is screened? From what I gather, this new health report was not driven by customer demand, nor were healthcare providers calling for more access to MUTYH testing. Generally our guidelines recommend MUTYH testing only in very specific circumstances, such as when a patient has multiple adenomatous polyps, or if they have a sibling who has been diagnosed with MAP. But it is not the healthcare providers or guidelines directing who is screened. Presently it would seem that 23andMe decides who gets screened for MUTYH variants, and with the blessing of the FDA. I am curious to know how my genetic counselor colleagues are handling or are likely to handle such cases. What would you most likely offer to a patient that comes for genetic counseling with a heterozygous positive DTC MUTYH variant report, and no family history concerning for MUTYH polyposis or other hereditary cancer syndrome?
My hunch is that most genetic counselors would offer follow-up clinical testing, and that this will usually be at a minimum a full analysis of MUTYH. I certainly can see the benefit of identifying patients who have biallelic MUTYH variants to allow for earlier and more frequent colonoscopy screening and prevention of some cases of early onset colon cancer for those who may not otherwise have known that they were at increased risk. Since the announcement of the new health reports by 23andMe I have been curious about how many individuals with MAP could be identified through this stepwise screening, first with 23andMe and then follow-up clinical testing of the entire MUTYH gene for all of those who first have one or two variants detected through 23andMe.
Staying with the estimated 4 million 23andMe customers of European descent who opted for the health reports, I estimate that there are about 390 total people out of the 4 million who have MAP, and of these about 250 would be identified through testing the two common variants alone (homozygous or compound heterozygous for Y179C/G396D. (My calculations and assumptions are all available here.) Through this stepwise screening of sequence analysis only for those who are first found to have at least one common variant, we would identify approximately 374 of the total 390 people with MAP, while missing about 16 people who carry two less common MUTYH variants. The cost per MAP case identified depends on the cost of follow-up genetic testing. Publicly available pricing suggests that single gene analysis of the MUTYH gene through a clinical lab is typically between $500-$1,500 per test for complete single gene analysis, although clinical testing may be priced as low as $250 to more than $2,000 per test. Thus, through this stepwise process (full gene analysis for those with one or more common variants first identified) the cost for clinical genetic testing would be somewhere on the order of $42,000 – $378,00 per MAP case identified. Given that general population guidelines recommend colon cancer screening should begin at age 50 and because approximately 75% of colon cancer in individuals with MAP occurs in patients under the age of 50, this genetic screening approach, and initiation of colonoscopy screening at an earlier age than typically recommended, could prevent as many as 94 cases of early onset colon cancer in this 23ndMe screened group, with cost per case prevented in the range of $168,000 to 1.5 million. It is important to note that this only considers genetic testing. A more complete economic analysis may include genetic counseling costs, additional cancer screening costs, as well as the estimated cost for genetic testing in other family members.
23andMe isn’t the only company that has an interest in MUTYH screening. Several labs offer DTC, or the close relative of DTC testing, the so called “patient-initiated, physician-mediated” testing in which an individual may request testing themselves, and a physician contracted by the lab signs off on the test order to bypass the need for FDA scrutiny (a discussion for another post). MyHeritage, who previously offered only ancestry testing recently began offering a “health” component which includes four variants in the MUTYH gene (including the two common European variants). Color Genomics and Invitae both consumer-initiated, physician-mediated tests as well as clinician ordered tests that provide a more complete analysis of MUTYH and other cancer susceptibility genes.
Is more comprehensive testing such as full gene sequencing better when it comes to MUTYH testing in the general population? It depends on how better is defined. I can certainly appreciate that detecting those additional 16 individuals with MAP (and being able to prevent about 4 cases of early onset colon cancers) that would have been missed in our 23andMe population of 4 million people would be one benefit to more comprehensive testing. But what’s the trade off?
At the recent American Society of Human Genetics meeting in Houston, Invitae presented data reviewing MUTYH variants found in analysis of a cohort of 270,806 patient samples that were submitted for multigene hereditary cancer risk panels. They reported that through a more complete analysis of MUTYH they discovered 5,929 patients with pathogenic or likely pathogenic variants in MUTYH, of which 1,377 (23%) of these patients carried one or more variants other than G396D and Y179C. That’s no surprise – in fact it fits well with our current understanding of the population frequencies for these genetic variants as well as our appreciation that about 80% of variants detected through MUTYH testing in a population of mostly individuals of European Caucasian ancestry are the G396D and Y179C variants.
The presented statistic that is getting the most attention in the news since the meeting is Invitae’s assertion that 40% of individuals in the cohort with biallelic mutations would have been missed by the targeted variant testing. But the necessary context that is missing here is that about 90% of those that would be “missed” in that 40% would have one of the two common variants paired with a second other, undetected variant. Through a stepwise approach in which only individuals with one or more of the common variants on first screening through 23andMe had follow-up clinical sequence analysis, approximately 96% of the total MAP cases would be identified. That means there would be only about 4% of individuals who would have biallelic variants that both would have gone undetected by a stepwise approach that begins with a two-variant screen.
If we were to ask Invitae who should be screened, I believe they would likely agree with 23andMe that everyone should be tested. Where they differ is in how the screening should be done, with Invitae advocating for more complete next generation sequence analysis for all people rather than targeted SNP panels. I can appreciate the logic in this argument that this strategy would detect more cases of MAP, especially in people who are not of European ancestry. But how would the cost of testing compare for the number of cases detected? The devil is in the details of how this testing is done and who pays for it. Invitae currently offers a patient-initiated, physician-mediated cancer gene panel for $250. If people elected this test and paid for it on their own instead of having 23andMe, given that it is considered a clinical test and no follow-up studies would be needed, then we would not see the additional cost in follow-up clinical testing (although we may see more downstream costs with follow-up for variants, including variants of unknown significance in other genes on the panel). My sense though is that people seek out tests like 23ndMe for reasons other than to know there MUTYH or other cancer gene status. So if we are considering the total cost of clinical testing for all people rather than just for those who present with at least one of the common mutations on a screening DTC test we would see quite a dramatic increase in the cost to identify a case of MAP as well as the cost to potentially prevent a case of colon cancer before the age of 50 (61-fold increase over the stepwise approach).
What is difficult to tease out in these analyses is that most tests that include MUTYH are not ordered with a high suspicion or concern for MAP. In most cases, it is included as part of a bigger panel, and MUTYH is not the primary reason for testing. In the case of 23andMe it comes along as part of a bigger ancestry, traits and health screen. In the case of Invitae it usually comes along as part of a multigene cancer panel for which the primary indication is usually for evaluation of genes other than MUTYH. So should we consider that extra information provided by the MUTYH analysis in patients a bonus or a burden? As DTC and consumer-initiated testing becomes more common, and as clinical testing panels more often include MUTYH, are we spending more time counseling patients about MUTYH results that are unlikely to significantly affect health outcomes or screening recommendations for the patient or their family members? I worry that patients may put too much value on these results as a possible explanation for their cancer, when it may unlikely be the case. And I feel conflicted about how much we should be advocating for family member testing when there is such a small chance that their care and outcomes will change as a result.
The focus right now seems to be on the question of how we should be screening for MUTYH. I think we are engaging in the wrong discussion here – I think we should go back to the heart of this and really take a close look at why we are doing this testing? What is driving adoption of expanding genetic testing to include analysis of genes such as MUTYH? And how should we integrate these technologies into practice in a way that will help the most people and cause the least harm. We need to critically consider the statistics that are presented and carefully weigh the costs versus the benefits of expanding testing.
—————————————————————————————————————————–
Given that we are just a week away from the annual NSGC conference, I’d like to make a plug for the 3rd annual ‘GC’s Got Talent’ hosted by Genetic Support Foundation on November 6th from 7-10 pm. The event will be Emceed by the ever engaging and talented DNA Exchange blogger, Bob Resta. In addition to storytelling, dancing, singing and comedic acts, a silent auction will showcase the artistic and creative talents of GCs. It’s not too late to sign up to share your talent and we are seeking storytellers for the themes: “Confessions of a Genetic Counselor” or “Most Awkward Moment in my Career”…Contact info@geneticsupport.org for additional info.
The DNA Exchange 
Thanks for writing this. To be fair, I’m not sure that 23andMe would say that MUTYH screening should be done by a two variant genotype panel. I think they might say that’s what they’re able to do in a way that’s cost-effective (doesn’t raise the cost of the test). Since MAP is recessive, the idea of population screening for it is is also unique since we wouldn’t necessarily pick people up for testing based on family history. Whereas for most hereditary cancer mutations, in a perfect world at least, we have family history (or knowledge of absence of fam hx info) that informs how we think about who to offer testing to. But as Katie said, this is all a pretty academic discussion because 23andme is making the decision. And the healthcare system is probably(?) absorbing downstream costs–and/or benefiting from any reduction in costs.
True Eleanor. I am not really sure what 23andMe would think is the best method for screening. One point I was hoping to make though was that targeted screening with a limited panel that includes variants for which are relatively common in a screened population and for which we have good data to support clinical implications, might make sense from a population health standpoint – in this example using the 23andMe published population frequencies, we would identify approximately 64% of those with MAP in the first pass, and a total of 96% who have MAP by starting with a targeted screen in a specific population, and only sequencing those who have at least one common variant. Of course 96% isn’t *all* cases and when we think of the implications of this screening for each individual it gets pretty complicated. I thought about that a lot about the individual implications when listening to your excellent podcast and interview with Stacey Wilson https://www.greygenetics.com/mutyh-map-and-colorectal-cancer/. And of course outside of those with primarily Northern European ancestry our knowledge of the genetics of MAP is much more limited. So to identify nearly all cases may be possible with sequencing, but at what cost? And what do we trade off? At what price point do the benefits outweigh the harms or the trade offs for population screening? These are questions that would be great to consider before embarking on mass population screening.
All great points. It’s also sad that the easy population screening (two common variants) would disproportionately benefit those of Northern European ancestry – possibly further increasing healthcare disparities. For the record, Grey Genetics has had zero patients with MUTYH findings on 23andme schedule an appt or even inquire about a consult. I wonder if other GCs in private practice are seeing these patients, if those with biallelic mutations at least are making it to GCs in hospitals – or if doctors are handling this. Or if people’s eyes are just glazing over the MUTYH results.
We are in the early days here. These MUTYH variants are increasingly being reported on families with no Colon ca hx from DTC “ancestry testing” as well as from extended panels aimed at families with breast and other cancers. Past evaluations have focused and are heavily prejudiced by families with Co ca hx, there may be more to this story yet (think about PMS2). I don’t see much discussion about the cost of cascade testing into the family, once a presumed pathogenic variant is found? While the impact to the index patient with the monoallelic MUTYH may be relatively small, this MUTYH could presage more problems elsewhere in the family……if the European carrier frequency approaches 1/100 or so, are people offering mutational status for the spouse or other first degree relatives? What about families featuring consanguinity? We may eventually find that MUTYH may participate as one of many factors in a multifactorial risk assessment much like what is going on with the WISDOM trial and breast cancer. The deeper we dig, the darker it gets!!!!! Good luck.
Carey Johnson MD Cancer Genetics and Genomics
Great read thaankyou