Thursday, September 3, 2009

Some Confusion Exists

I have a great comment string going on with Daniel MacArthur over at his blog Genetic Future

I think there is some confusion going on here and I place blame on just about everyone in this space who has a mouthpiece.......

But mainly I lay blame on the marketing teams for the Direct to Consumer Genomics companies.

These companies have an interest in making you "think" that their products have some particular health relevance.

Otherwise, no one in their right mind would waste their time with these tests.......Other than the HUGE field of ancestry buffs like Blaine Bettinger J.D. (woohoo)
We need clarity here.

From Daniel- "The American College of Medical Genetics is saying "Genetic tests of individuals or families for the presence of or susceptibility to disease are medical tests."

The fine print says:

"This guideline should not be considered inclusive of all proper procedures and tests or exclusive of other procedures and tests that are reasonably directed to obtaining the same results. In determining the propriety of any specific procedure or test, the geneticist should apply his or her own professional judgment to the specific clinical circumstances presented by the individual patient or specimen"


"The judgement of what constitutes an inclusive test is left up to the physician"

In my opinion, there are Green tests, Yellow Tests and Red Tests. I think Ryan Phelan sized this up pretty eloquently in 2007 at a conference I spoke at with her.

Red means - Stop, does this test need analytic/clinical utility/validity? Yes, Go ahead and regulate, these are clinically validated/used clinically for a long time, tests which have a use in medicine. If this test is claiming to do so but does not, then this too should be a cause of regulation.

Yellow means - Well, this could be used but maybe hasn't yet. I think of Age Related Macular Degeneration testing. They could have clinical applicability and haven't been put to use yet. This category may also include low odds ratio common SNPs here which indicate risk FOR disease.

You should have some caution when selling/regulating these tests. The biggest problem is that with evidence evolving over time the low ORs may actually be overturned or fall into the Red category.

They key point is what the test claims to do here.

Does it claim to tell you a risk for a disease via algorithm etc.?

If yes, then it becomes a Red test.

If it says in huge disclaimers, this test DOES NOT PREDICT DISEASE RISK, in plain and clear writing on every piece of its marketing then Yellow tests stay Yellow.

Green Tests - Have absolutely nothing to do with a person's health. These tests do not need medical regulations. Ancestry could be here. Eye color/ear wax/height.

But the moment it is used for medicine or medical procedures (PGD for these things, scary but possible) it then becomes a Red test.

ACMG is talking about Red tests. It is also saying, if you as a clinician think a Yellow test is actually a Red test, then it is a Red test.

Which I agree is confusing. But ACMG is not the US or State Governments. Nor is it the UK or anywhere else's government......It is a professional organization.

My take is simple. If there is a risk for public harm, the government should protect its citizens from that harm in a reasonable manner. I emphasize, reasonable manner. What Techies in the Silicon Valley view as unreasonable regulation may very well be extremely reasonable in the view of physicians and hospitals......

There is no reason to get all crazy here. The New York Times is right, until the government or state governments step in to protect the citizens of risk,
it IS Buyer Beware when it comes to genetic tests.....

In my mind, genetic tests need to prove their worth in the field of medicine. They do this by medical science and clinical science studies.

Without these, they are useless noise.

So, should you be able to buy useless noise?

Go ahead.

But the moment the noise whispers in your ear "PSSST, I can tell you your risk for diabetes for 50 dollars" Its A$$ should be regulated.

I hope that clears things up. The problem here is that you have a company selling you noise mixed with clinically valid tests. They are the first company whose A$$ should be told to hold up and split the products.

Then you have a company selling you noise all the while "inferring" it is actually predictive and useful in medicine by getting Doctors to use it in medicine should also be regulated.

Blame Marketers and Spin Men, just like "Thank You for Smoking"

The Sherpa Says: If you apply the Red, Yellow, Green interpretation scale you will soon understand why all the confusion exists. And how that confusion can be cleared up.


Anonymous said...

Dear Sherpa,

The only confusion here is in your mind. Listen: doctors like you have nothing to do against IT and genetic technology. Your future as na doctor is DARK, as machines and IT will sustitute people like you.


Steve Murphy MD said...

LOL. They couldn't even get rid of us in Idiocracy. Dream on earthling and hope you don't get sick until the super secret project you are working on goes Beta.........

BTW, "this one goes in the mouth" OOPS.

-Dr Lexus

cariaso said...

While the specific terminology of Red/Yellow/Green may be new, the idea of "approved snps" is not.

What is not clear is who should make that call. It seems there is lots of vague finger pointing but few volunteers.

The Coriell ICOB has approved a whopping 5 snps, while their comments suggest they'd like to see NIH/CDC or some other authority provide that guidance. Any snp not on their list of 5 is not released. I consider this a good example of snps which meet your RED definition. Do you feel that all 5 of these snps

should be considered Red?

23andMe has decided that perhaps 150 (my own guestimate) snps are worthy of communicating via their reports.

Are their any in their current reports which you feel are being released, but which should not be, as they do not meet a sufficient standard?

Do you have any objections to their release of *all* raw data to customers? Are you asking for Red snps to also be removed from the raw data?

When a snp has not yet been classified as Red/Yellow/Green what default should be assuemd Red or Green?

If you have any specific examples of snps which you feel belong in a particular category I'd be extremely interested in their rs#s.

Steve Murphy MD said...

Great questions.
I will get back to you on the criteria for each SNP.

As you may not be aware. I am Pro-Access to raw data. Hell, everyone should have access to their raw bioinformatics. But the interpretation is what should be regulated.

I have always maintained that the interpretation of disease related SNPs is Red or Yellow. Disease related SNPs that are used to predict disease risk are Red.

There are committees like the one I sit on for CPMC. We have voted on more than 5 SNPs BTW and Coriell plans on releasing others soon.

EGAPP is another good group for this work. Unfortunately, each DTC company has established some of their own criteria. Which include silly things like "At least 250 people were subjects of the research"

That is hardly a useful criterion.

Sites like yours which are transparent and give the science SHOULD be differentiated from those companies which do tests AND interpret those risks.

Raw data is ok. Intepretation is the key here. And interpretation of biologic samples from humans for the goal of predicting risk of disease is medicine, unless it is research which in that case is Medical Research. All of which merit regulation.


Anonymous said...

Dear Sherpa

We need you on twitter to ensure your thoughts are reaching the ever growing group following the personalised medicine discussion at #personalized medicine. You can tweet a link each time you update your blog and draw more people into your discussions not just dgmacarthur!!



Steve Murphy MD said...

Ah Hah. Check twitter. I am on it!

Anonymous said...

Does Promethease provide "interpretation" of raw data? The output sure looks like it comes from an algorithm and it speaks frequently of risk based on SNP associations, some of which may be quite sketchy. If interpretation is what should be regulated, then Promethease should be regulated as a test/device/assay and SNPedia is practicing medicine by promoting Promethease.

Or is it "claim" that should be regulated? In this case, SNPedia does not claim to provide health risks, so it should not be regulated. However, Promethease is still interpretation, so the lack of a claim here is equivalent to DTC companies' disclaimers of being for educational/info purposes while still providing interpretation.

This is an important question, because the heart of your argument against DTC companies is that their substance (= interpretation) is what should be regulated, no matter what the form (= claim). If this is really your argument, then SNPedia and Promethease should be regulated.

If your argument is that form is what matters, then neither SNPedia/Promethease nor DTC companies should not be regulated.

Steve Murphy MD said...

Interpretation should be regulated. But the definition of regulation has yet to be determined.

DTC Genomics is NOT Prometheus. Prometheus does not obtain human biologic samples.

Obtaining samples, testing and interpreting is the practice of medicine.

Prometheus is NOT medicine.

See my post on 4 September 2009.

Maybe the regulation should be to subject your software to a third party to determine if it is "accurate"

Otherwise you will have all the crazies tell you your hidden "traits" based on your raw data.