Sunday, November 28, 2010

Personal Genetics & Utility: Round 2 – Mind the EGAPP

ResearchBlogging.org
Yesterday I wrote about the false Family History vs. Personal Genetics battle, today I look at the old chestnut of traditional risk factors. There seems to be a lot of fear among some professions that personal genetics is attempting to take over their jobs – it’s been like this from the beginning mainly due to misunderstanding (wilful or otherwise) exactly what personal genetics is and what it’s role in healthcare can be.


The latest salvo is from the EGAPP Working group who published their assessment of genetics vs. traditional risk factors (TRF) in cardiovascular disease risk. They looked at the 9p21 variant as well as 57 other variants in 28 genes associated with CVD and they sought to document

“the extent to which genomic profiling alters CVD risk estimation, alone and in combination with traditional risk factors, and the extent to which risk reclassification improves health outcomes”.
Some conclusions from EGAPP:
  • The magnitude of net health benefit from use of any of these tests alone or in combination is negligible.
  • The EWG discourages clinical use unless further evidence supports improved clinical outcomes.
  • the overall certainty of net health benefit is deemed “Low.”
  • the estimated additional benefit from adding genomic markers to traditional risk factors was found to be negligible.
  • Traditional risk factors such as those used in the Framingham Risk Scores have an advantage in clinical screening and risk assessment strategies because they measure the actual targets for therapy
  • To add value, genomic testing should lead to better outcomes than those achievable by assessment and treatment of traditional risk factors alone.
  • To be useful, genomic testing should provide demonstrable improvement on the predictive value of TRFs.

Fine, fine, fine, all correct and proven, but all missing the point completely. It does not matter that the genetics did not add anything, even with the legendary 9p21 variant. Why should personal genetics be thought of as a replacement for traditional risk factors? EGAPP in it’s narrow scope is correct, but as usual the “negligible benefits” etc. will be, actually are being, quoted widely to trivialise personal genetics, just as the family history study was used to consign genetics to irrelevance.


The world moves on and nothing changes. In the early days, almost 10 years ago, it was the same, the genetic risk had to be “over and above” traditional risk factors. But why? What is expected of genes, are they supposed to possess some transcendent quality so that some sort of independent risk factor emerges from a genetic profile? Or is it that genes code for proteins that function in the various pathways, the perturbation of which can lead to metabolic problems (the traditional risk factors) and eventually disease?


I get told off for car metaphors but here goes. Driving along in the rain, hit the brakes, skid, crash. Skidding is a risk factor for crashing, I can try to reverse the skid, it might work, but I would rather avoid it in the first place by driving better in the rain (at least until pharma comes up with the anti-lock brakes pill).

The aim of personal genetics is to prolong health. High blood pressure, low bone mineral density, arterial plaques, etc., are not present in healthy people. They might be useful indicators in predicting disease, they might be useful values to put into the Framingham calculator, but they are best avoided in the first place.


All this is obvious – so why is it that genetics is compared so frequently to classical risk factors? It’s not a surprise that they don’t contribute more, why should they? Genetic variation does not have this magic “over and above” quality. But it is there from birth, it is there even in healthy people. This was mentioned in an earlier influential paper comparing SNPs & risk factors in type 2 diabetes, the authors claimed that the genetics added little to nothing but did add as a by the way:

“Although genetic information appeared to be useful when only factors known in youth were considered, genetic information in the context of risk factors measured in adulthood did not help to refine the prediction of diabetes risk” (Meigs et al)

 

So it was only useful in youth, in healthy people. Dammit. When people are actually ill the traditional risk factors win hands down. Of course. So the contribution of genetics is “negligible”. No use for genetics in healthcare yet.
But what about the poor healthy people who want to stay that way? No family history for anything I particular. Normal BMI, fat mass, lipids, blood pressure, glucose, insulin, HbA1c, and so on… Will regular medicine and TRF testing still win? No, it can’t. I’m not saying that genetics will definitely win, but it’s certainly favourite, at least it has the possibility of scoring where regular medicine does not.


EGAPP is probably correct, not ready for use in the clinic – at least not in the clinics that most of us are familiar with. But this is not the same as not ready for use. Most doctors I have spoken to (many) want genetics to be a simple test that classifies risk; high, medium or low, and tells them what to do. This is a reasonable desire and fits in with the way most of them work – a few minutes per patient, clear decision making advice required. They have no time for a long interpretation and explanation of small risk changes, up or down, how to ameliorate raised risks in the long term through diet & lifestyle, etc. It’s not their job, mostly. But if “CVD is a public health care concern” (EGAPP) it needs a public health care approach and if genetics is going to be involved it will not be as a replacement for conventional risk factors but will be incorporated into healthcare long before conventional risk factors even begin to raise their ugly heads.

This is where the research should be going: proper assessments of personal genetics vs. standard health information with healthy people. Is genetics better than classical risk factors in healthy people? (of course this is just the same as asking “is genetics better than NOTHING” – which is exactly the right question). I’m expecting that 23andme will be exploring just this – they have the money, the skills, the database and the experience, and of course some business interest (as everybody does, including MDs and clinical geneticists). But I hope EGAPP will do it too.


It’s not genetics vs. regular healthcare, it’s when, where and how to use genetics in healthcare.

PS – The PHG foundation has a nice report on the Genetics and Public Policy Centre survey.

“A random sample of 1,048 US customers of the three major companies offering personal genomics DTC (23andMe, deCODEme and Navigenics) were surveyed online between June 2009 and March 2010…58% said they learned information that would help improve their health, and as a result of testing, 34% said they were being more careful about their diet and 14% were exercising more…This study provides long-overdue evidence that consumers are satisfied with DTC genetic testing services, and are generally able to interpret their results. It also indicates that there may be direct health benefits resulting from the tests in terms of behaviour modification…this survey indicates not only the absence of harm caused by DTC genetic testing services, but also the possibility of benefits.”

 

Early days, these were early adopters driven by some reason to take the tests, but a promising start.

Evaluation of Genomic Applications in Practice and Prevention (EGAPP) Working Group (2010). Recommendations from the EGAPP Working Group: Genomic profiling to assess cardiovascular risk to improve cardiovascular health. Genetics in medicine : official journal of the American College of Medical Genetics PMID: 21042222

8 comments:

  1. So why not just lie and tell everyone they have this "mysterious" Genetic risk? Placebo effect for behavioral modification. Just like Santa Claus, Easter Bunny, etc? Seems a lot more cost effective.

    ReplyDelete
  2. Adding to Keith's points, it's really too early even to be having this sort of discussion. Why? Because the real value of personal genomics will not arise until the full genotype of an individual becomes available. Sure, it will take several years for more fully-predictive info to become available (decades for a $1M interpretation?), but there will be significant value available well-before that in terms of personal health understanding and guidance. Personal genomics really needs to be placed in the proper context, as Keith mentioned, and he just gave two examples of how that is not happening. IMO, ANY personal genomic info available to me is a far better option than no information at all. Who wants to regress to 20th century healthcare where genes were seldom even discussed?

    ReplyDelete
  3. @Robert
    Noise is better than absence of noise?
    Just because it's there doesn't mean it has a use.
    This argument you and Keith make is kind of like saying, "So now we have an MRI machine, why not just run everyone through the tube to see what our basal ganglia look like?"

    ReplyDelete
  4. Steve - no obviously. But how do you know it is noise? We don't know for sure and certainly the types of studies comparing genetics to traditional risk factors and family history will not give us the answer.

    The point, that you have missed, of the arguments presented here by myself and Robert, is not that it must be used now, but what is the correct use and how to test it (note in Bob's comment the use of phrases like "too early", "several years", "will be").

    It's no good saying that because it doesn't add anything to traditional risk factors it has negligible value. That type of study does not add anything to our knowledge and has neglible value. It shows a misunderstanding of the potential uses of personal genetics which runs very deep in the medical and clinical genetics communuty.

    If we do a thousand more studies like EGAPP we will still not be any wiser - I can predict right now that if a patient presents with high blood pressure and glucose intolerance it would be better for the doctor to treat those symptoms rather than ignore them because they don't appear in his personal genetics profile.

    ReplyDelete
  5. @Steve

    Noise, yes, for some, but not all, and only temporarily for most (but likely not all). My e3/e3 status is not noise, though it is not clinically actionable either, though if I was instead e4/e4, I would be much more vigilant about my future actions.

    Also, you need to bring "carrier status" into your inclusion list, since those data are just as valuable as drug sensitivity, especially to the younger generation who have not yet married. Those data we do not need to wait for, for a large number of diseases.

    Clearly, the prevailing evidence indicates you respect the value of personal genomics as much as Keith and I, and your devil's advocate approach is both NECESSARY and fully appreciated.

    -Bob

    ReplyDelete
  6. @robert,
    Thanks. I do see great power here. I don't want it to get watered down by shennanigans which fail to produce (yet amazingly build !BUZZ! Whatever that is) and hence water down the trust and power.

    I agree, carrier data and PGx data are key. As I deliver a bunch of Counsyl reports this week to patients.......

    I believe, I just don't drink the Kool-Aid

    -Steve
    p.s. thanks for noticing.

    ReplyDelete
  7. @Keith,

    Here are the studies that need to be done

    1) Sham Genomic scan vs True Genomic scan and behavioral modification results, cohorted over 10 years

    2) Sham vs True and diabetes/CVD/HTN incidence

    3) Genome Phenome Metabolome Cohort over 20 years

    4) Whole genome of disease trios and subtraction analysis from "normals"

    5) the 100,000 genome project with 10 years of cohort data

    None of these will have results available for 10 years at least.

    The longer we wait, the longer we wait.

    Trust me, the genetics community is not medicalizing this anymore than it need be.

    But it is useful to ask if this stuff is any better than the palm reader in Time Square.

    -Steve

    ReplyDelete
  8. @Steve

    BTW I agree with Bob about the Devil's Advocate, especially in the absence of standards/oversight/codes of practice.

    I pretty much disagree with your list though. How much of that type of research has been done for a) Family History b) Traditional risk factor algorithms c) Warfarin/Plavix and d) BRCA???

    Again a different standard is being required of personal genetics. There are also real ethical difficulties with giving people long term advice based on sham genetic results. Also because of the complexity of the systems under study even after 10 years the results will not give clarity. Let's face it, we don't even have "proof" that smoking causes cancer - it's never been tested with sham cigarettes. Might sound absurd but I could argue that lung cancer could be caused by raising my hand to my mouth regularly for 5 minutes 40x per day. Or by inhaling in a particular way, etc.

    The context of personal genetics is prevention. If a sick person comes to see you with BMI=35 and glucose all over the place is it clinically useful to consult the patients genetics to see if he/she is at raised risk for obesity and diabetes? Of course not.

    Personal genetics will (note: will) have it's true value in preventative healthcare which may or may not involve a healthcare professional. What is the “medicine” of preventative healthcare? It’s not aspirin or statins, it’s diet, behaviour and lifestyle. When you advise your patients on a healthy diet not a single piece of the advice that you will give will have been tested using clinical trials.

    We don’t need to wait for the results of the trials you propose, we all know really that when faced with a symptomatic patient that there is not much use for genetics unless a) it’s a rare hereditary cancer or other disease or b) for selecting/dosing medication. There’s not even a great deal of direct use for most of the risk panels in preventative healthcare – the advice becomes too general. There are benefits for those who will use the information to learn more about prevention and will be motivated to make changes, and will understand more about why those changes are necessary.

    Not ready for prime time does not mean not ready – I interpret the former as meaning not ready for universal inclusion in public healthcare. It will be obvious when it is ready, when the benefits easily outweigh the potential risks – 10 year clinical trials will not be needed (nor would they be effective).

    An example: Homocysteine and MTHFR. With genotype 677TT and a low folate diet it is almost certain that homocysteine will be raised and that this is easily reversed by ensuring adequate vitamin levels in the diet (at levels way below the upper limits). Not all agree that reducing homocysteine is a necessity, there is no final proof that it causes strokes etc. But there is a mass of evidence from observation and epidemiology that it is a risk factor, while the prospect of harm of having adequate B vitamins in the diet is minimal. So I might as well keep my levels low, it’s not hard, it’s no guarantee that I won’t have a stroke but it’s a good guarantee that I won’t get a stroke caused by high homocysteine.

    Here’s the problem, personal genetics is not going to revolutionise medicine, it’s not going to provide magic routes to eternal youth, it’s going to be pretty boring in it’s implementation, and it’s benefits will be seen over years and decades, not days or weeks.

    ReplyDelete