Thursday, November 20, 2008

Genes provide Minimal Benefit!!!! Seriously Oprah!


Genotype Score in Addition to Common Risk Factors for Prediction of Type 2 Diabetes......

That's right, the New England Journal of Medicine has 2 publications in the Journal this week. What's the take away?

Paper One: Conclusions A genotype score based on 18 risk alleles predicted new cases of diabetes in the community but provided only a slightly better prediction of risk than knowledge of common risk factors alone. The C-statistic was 0.900 versus 0.901......some on....can we even call that better? The nearer the C statistic to 1, the better the predictive value of the test...so can we really say it is worth it???

Paper Two: Conclusions As compared with clinical risk factors alone, common genetic variants associated with the risk of diabetes had a small effect on the ability to predict the future development of type 2 diabetes. The value of genetic factors increased with an increasing duration of follow-up. In this study they use the area under the ROC, where the change is 0.74 to 0.75....again lackluster results...

You tell me? I just gave a lecture to several residents in Internal Medicine......I asked what the barriers where to taking a family history. One pragmatic resident said "So let's say I get the family history. What will I do clinically different?"

These articles in the New England Journal of Medicine make me pose the same question.......

What do I do with this only slightly, increased ability to pick up disease risk?

The answer? No one knows......Is it worth spending thousands of dollars on a test that doesn't dramatically improve detection of predisease risk?

I am not so sure.....

The bigger question is, are predisposition SNPs what we should be studying instead of rare copy number variants????

The Sherpa Says: With stats like this, DTC companies will be dead and so will the microarray business.....Will we achieve personalized medicine with such disappointing results???? Yes, but not via DTC...or via Oprah.

6 comments:

Daniel said...

Oh, come on Steve. Firstly, common SNPs are just the beginning, and personal genomics companies are well aware of that - that's why they're already starting to offer sequencing services, which we'll certainly see increase in scale as sequencing costs go down. The future is in rare variants with higher odds ratios: we'll see a host of publications on these over the next couple of years, they will certainly provide far more power for individual risk prediction, and personal genomics companies will be adding them to their chips as soon as they've been validated.

Secondly: with the rather lame exception of Navigenics, consumer genomics companies offer more than just disease risk estimates. People are just as interested in the information on ancestry and non-disease traits as they are in disease risk, if not more so - see Blaine's post today confirming at least 450,000 customers for Family Tree DNA. The low clinical utility of these early-phase disease markers will not kill the personal genomics market.

As for the microarray business - it will persist for the next few years on the strength of targeted diagnostic assays. Next-gen sequencing is hurting the chip market, but it won't kill it for a long time yet.

Steve Murphy MD said...

I agree, that's why I have reached out to Geni.com.....

But here's the dilemma....as you approach real ORs you get awfully close to a clinical diagnostic test.

Ancestry combined with medical history is a powerful tool....combined with rare variants it is what will enable GENOMIC medicine....

The em PHA sis on the wrong sil AH ble....it should be genomic MEDICINE.

-Steve

Anonymous said...

Steve, please don't missinform the readers.

From paper 1, the C statistic was 0.534 without the genotype score and 0.581 with the score (P=<0.01). This mean that clinical data together with genetic data (only 18 markers) is significantly more accurate to predict T2D than clinical data alone.

From paper 2, the accuracy on the prediction of T2D using both clinical and genetic data was significantly much better (P<0.001) than using clinical data alone. By contrast, the accuracy of clinical data alone was decreased with age.

Taken together, these important results, published in the most prestigious medical journal, NEJM, suggest the utility of genetic data to help in the prediction (and prevention) of T2D.

Steve, please don't missinform the readers. I guess you have some conflict of interest to missinform the readers. Upps, I forgot, maybe are you starting a company with a diferent business model than DTCs?

Steve Murphy MD said...

you sir are a manipulator...read the paper. I did and reported the most important results....those controlled for nearly every variable possible.....as for conflict...the only conflict is between good and bad science....DTC is bad science....as you demonstrated....

-Steve

Anonymous said...

I agree with anonymous. I think you Steve are the only manipulator here. You are chronically obssesed against DTCs. You report just the things you think benefit you. That's bad science my friend.

Why do you do this? Because DTCs are getting the market you would like to get for you. Because they are making money and you don't. Because you are jealous of them.

Good luck! You will need it...

Steve Murphy MD said...

@ Anonymous,
I am sorry you think I am a manipulator. I reported on the accurate and best controlled statistics and provided a link to each study.....If that is manipulation....well, you are mistaken.

Are DTC companies making money? I wouldn't know. Am I doctor doing medicine? Yes....isn't that a little different than selling tests??? I think so......

I am not jealous of someone who has the government looking to regulate them further and who has a test which is not clinically relevant. I am proud of them for being "out there"...but I am very different than them...

Why hate on the doctors????

-Steve