Friday, November 5, 2010

Family History Better than Navigenics/DTCG Shill for Cancer Genes?

Yes,

You heard it here. A recent study abstract and pressed about from my friend Charis Eng MD PhD, Clinical Geneticist, Internist and all around really smart lady spoke today about her findings of a head to head, DTCG vs Family History at discovering cancer risk.
I actually sent some data Ken Offit's way about a similar thing way back when, Ken is yet another, really smart guy. He wasn't surprised. Nor was I when I heard Dr. Eng's findings.

First, Caveat Emptor

This is an abstract! Repeat after me......

What does that mean?

1. It is not peer reviewed fully
2. It is not published yet
3. It is preliminary data

This test was Navigenics Compass vs Family History in 22 females with breast cancer, 22 males with prostate cancer and 44 people with colorectal cancer.

What was the result? Family History placed far more people in the proper high risk category. 8:1

Family History put 22 people in the appropriate High Risk Hereditary Category, DTCG only one.


Further, it looks to me that the Navi "Gene" Scan missed several high risk patients who actually had MMR mutations (I.E. Genetic Cancer).....D'Oh.

First off, this is like a case study. But it signals a HUGE shortcoming of DTCG. False reassurance.

I have been beating this over the head for 3 years now! These tests that have "medical" relevance need to be couched with proper medical guidance.




There are huge shortcomings in the current offering of DTCG tests and those offering medical information need to be regulated as medical. This is a classic case in point of potential and REAL missed cases.

Not Good.

That being said. It is November AKA Family History Month. You should absolutely take your family history and bring it to your doctor. If they don't know what to do with it, call us. We do.

The Sherpa Says: No surprises Charis, I saw this with some DTC cases I have had, passed it on to Ken who passed it on to NIH and The IOM. This is the huge problem with hype and over promise. It always fails to deliver, unfortunately in this case at a great risk to consumers.

8 comments:

RodPhilp said...

I totally agree.. DTCG needs to be put into clinical perspective by a professional ... but there is a lack of qualified professionals.. especially in states like Iowa (where i live). On a brighter note - you may be pleased to know that this is one of the aspects that is being taught to first year medical students at our University (Des Moines University) ... which is a DO school ... not generally credited with being at the 'rock face' when it comes to genetics.

Jack Soda said...

Of course people who have put proper thought into this should not be surprised. What will be interesting is to see how quickly Navigenics, et al. mobilize to refute this. If they naively use arguments that the science is incomplete, those very words will very likely come back to very much haunt them.

Keith Grimaldi said...

FH is useful, of course but why does it have to be one vs. the other. DTC genetics will only give false reassurance if interpreted badly.

There are other caveats in this study:

a) It looks like it was a specific highly hereditary form of colon cancer

b) The family history method appears to have been a very detailed process, not just simply ticking a few boxes. A process that is not so easy and is time consuming

What I would really like to see is a study of FH vs. genetics vs. FH + Genetics for the common diseases where the genetic components are not so clear cut as in this particular study.

Steven Murphy MD said...

@Keith
OK. Here's where it really breaks down.

1. single SNP markers and EVEN multi plex testing has proven as poor prognosticators when put up against current clinical risk factors.

It is obviously early days for these types of studies, but there is a clear trend of general uselessness for clinical decision making for risk prediction.

2. 2C19 testing is obviously better than famhx for prediciting plavix response and DTCG has FamHx beat on that mark.

But I emphasize, this stuff (PGx, Medical traits) needs to be regulated if given via DTC. Why? If people will use these and take these to the doctors for making decisions, they need to be.

3. FamHx plus genetics may have some promise if NOT added to current risk markers like BP, HGBA1c, HsCRP.

I think of genes as just another biomarker with less accuracy than current phenotypic protein markers for common disease risks.

If I really want to know risk, I do a stress test and Reynolds risk for CVD.

I think the truth was told in the Consumer review, this type of test is "cool" for curiosity, but not the best for medical use overall.

That being said, PGx and Cancer genetics are not 2 fields that DTCG should be playing with. These should really be clinical medicine.

IMHO.

and the FDA's too.

-Steve

Keith Grimaldi said...

Hang on a minute, the subject is changing, we're not talking about genetics vs. classical risk markers (or regulation). The comparison is genetics vs. family history. Surely classical biomarkers in general are better than both - if there are raised risk factors there IS something wrong, may not be cancer or heart disease yet, but still, something is wrong. I don't want to get raised risk factors - I want to find out before how to prevent them. I don't want just to prevent disease but to prevent the pre-disease state too.

The continual comparison of genetics vs. classical risk factors misses the point completely - in one paper it stated:

“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, http://bit.ly/amewN4)

This was just a “by the way” mention in the last paragraph of the article – but surely that is the whole point of genetics isn’t it? Comparison with classical risk factors is a waste of time. What’s a better risk predictor of bone fracture, genetics or very low bone mineral density? Certainly the latter – great, but I would rather just have the former and never arrive at the latter. Skidding in the rain raises the risk of crashing – I can try and reverse the skid but I’d rather avoid both the skidding and the crashing.

It seems to be convenient to bash genetics because it does not perform better than classic risk prediction based on biomarkers - but that’s not a useful comparison. A useful tool is something that predicts the future before it happens, obvious of course, and raised blood pressure, or other risk markers, should be regarded as “happened”. The questions are:

1. Is FH better than genetics. This was the subject here and the study mentioned was not a fair setting for the comparison

2. Is FH + genetics better than either used alone?

3. 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 agree with you on the “genes as just another biomarker with less accuracy than current phenotypic protein markers for common disease risks”. I think that is exactly what they are, how they should be considered and how they should be tested. Less accurate? Probably. But are they more accurate than nothing?

I’m not making a case FOR genetics here, I suspect that in general you are right that the current situation the panels are not incredibly useful in the clinic, as far as strict clinical utility is concerned – the risk changes are small and the interventions are too general, almost everyone will end up with the same generic advice for a healthy life. But I will make a case for personal genetics having personal utility and a potentially valuable role to play in prevention – where prevention means preventing low bone density, high blood pressure, glucose intolerance, etc.

Regulations, DTC, MD, etc – all secondary questions, important, but secondary

Keith Grimaldi said...

Hang on a minute, the subject is changing, we're not talking about genetics vs. classical risk markers (or regulation). The comparison is genetics vs. family history. Surely classical biomarkers in general are better than both - if there are raised risk factors there IS something wrong, may not be cancer or heart disease yet, but still, something is wrong. I don't want to get raised risk factors - I want to find out before how to prevent them. I don't want just to prevent disease but to prevent the pre-disease state too.

The continual comparison of genetics vs. classical risk factors misses the point completely - in one paper it stated:

“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, http://bit.ly/amewN4)

This was just a “by the way” mention in the last paragraph of the article – but surely that is the whole point of genetics isn’t it? Comparison with classical risk factors is a waste of time. What’s a better risk predictor of bone fracture, genetics or very low bone mineral density? Certainly the latter – great, but I would rather just have the former and never arrive at the latter. Skidding in the rain raises the risk of crashing – I can try and reverse the skid but I’d rather avoid both the skidding and the crashing.

It seems to be convenient to bash genetics because it does not perform better than classic risk prediction based on biomarkers - but that’s not a useful comparison. A useful tool is something that predicts the future before it happens, obvious of course, and raised blood pressure, or other risk markers, should be regarded as “happened”. The questions are:

1. Is FH better than genetics. This was the subject here and the study mentioned was not a fair setting for the comparison

2. Is FH + genetics better than either used alone?

3. 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 agree with you on the “genes as just another biomarker with less accuracy than current phenotypic protein markers for common disease risks”. I think that is exactly what they are, how they should be considered and how they should be tested. Less accurate? Probably. But are they more accurate than nothing?

I’m not making a case FOR genetics here, I suspect that in general you are right that the current situation the panels are not incredibly useful in the clinic, as far as strict clinical utility is concerned – the risk changes are small and the interventions are too general, almost everyone will end up with the same generic advice for a healthy life. But I will make a case for personal genetics having personal utility and a potentially valuable role to play in prevention – where prevention means preventing low bone density, high blood pressure, glucose intolerance, etc.

Regulations, DTC, MD, etc – all secondary questions, important, but secondary

Steven Murphy MD said...

@Keith,
So let me get this straight. Something is better than nothing?

Even if it has a hint of a chance of gving increased risk it is good?

This makes me think back to the whole birth order business. Or maybe the phrenology days?

I agree, there are some genetic tests taht are useful for prediction, but even more importantly....we WILL find the missing heritability.

And THAT Day, THAT day we will have a thing of beauty. Unless of course some schmuck keeps pumping money in the "Next Big Thing" only to be proven time and time again they are incorrect.

This type of thing will erode the public and the healthcare provider's trust in the field.

This is and continues to be what is happening.

Is a 9p21.3 SNP, good enough?

Will it help you lose weight better than a kick in the ass from your physician?

Will it help better than 3 appointments with a physical trainer?

What about 3 appointments with a nurse health coach?

If genotyping is free, this questin is moot. If not, this is the valid point here.

What is the value of something that only gives you risk akin to looking through the key hole of a door?

For now, to me, it is just noise.

You should NEVER use these tests to assess cancer risk. EVER!

-Steve

Keith Grimaldi said...

No Steve I'm not saying that, if that's the way it comes across then I need to clarify.

It's a real question, not rhetorical. The test should be is genetics better than nothing?

I'm not saying that just because there are small risk increases with some panels that it therefore must be better. The answer depends on what is better, i.e. the utility. As I said before I don’t think that the various panels will have a lot of clinical utility for the majority, everyone will have higher risk of something and it is clearly not “better” if every healthy person rushes off to the doctor for help. On the other hand, for personal use, it could be “better than nothing” to have this genetic/health information as long as the context is clear, that the risks are small, they do not indicate immediate danger, they are estimates based on current research, etc, etc. The recent survey indicates that this is the way they are taken, at least by the 1,048 people who responded (http://bit.ly/bUNKFF).

It’s easy to set up a study to compare genetics to classical risk markers, doesn’t take long and will get you a nice orthodox publication in NEJM or JAMA. It’s much harder to do the study to identify and quantify the benefits of genetics vs. generic information in healthy people, but this is where the real value lies.