Thursday, July 9, 2009

Ahh CRP, Genotyping is such sweet superstition!


Recently in the Journal of the American Medical Association an article Entitled Genetic Loci Associated with CRP levels and Risk of Coronary Heart Disease...... Long name, useful article.......

Why so useful?

Because it proves quite a point.

What is that point?

In most genetic associations the causal relationships are not proven. These studies are merely associations.....
Much like what I like to call superstition...... Perhaps you have heard of some of these?
If you spill salt and then throw it over your opposite shoulder you will avoid bad luck


If you break a mirror you have seven years of bad luck


If the cows are sitting down, it's going to rain.


Don't leave a fan on in a closed room or it will suffocate the occupants

Don't allow a cat to be in the room with a baby or it will suffocate the baby


I could go on here, but I think you get the point. There are millions of observations in this world which are so strong that they get carried on through folklore as if they are correct associations.

I am concerned that this genetic exceptionalism which is in the press and with the DTC genomics companies may create a whole new level of superstition........


"If you have SNP x then you will get disease Y............some of the time"

And it will be self fulfilling......trust me, you will see cases of this coming true being published as anecdotal reports, which will only strengthen this superstition about SNPs and maybe even other genetic findings like CNVs or other rare changes.....


This is why I laugh with associations without biology worked out.......

This is precisely what our ancestors did when the kept the sword which wounded them in a cool place to reduce the inflammation of the wound.......


Frankly, I think that using SNPs in this fashion is the EXACT SAME THING.......


So these people who are considered early adopters and are "cutting edge" here with 23andSergey scans or CEOs of DTC companies are actually nothing more than CAVE DWELLERS.......

This study in JAMA shows a smiliar finding. CRP also known as C-reactive protein is associated with increased risk of heart disease.............there are certain SNPs which are also associated with elevated CRP levels.....so using rat brain thinking, we postulate that the SNPs must ALSO be linked with increased risk of heart disease....... WRONG!!!!

rs6700896, rs4537545, rs7553007, rs1183910 and rs4420638 contribute nothing to risk for heart disease, but are significantly linked with elevated CRP.....

Which goes to show that there is a problem which cannot be solved by our current mode of thinking. Because there is a problem with the way in which we think.

And those who huddle together telling stories about there genome scans are nothing more than cave people exchanging superstitions.........

The Sherpa Says: Stick to the science, look for biological explanations and then look for clinical application. Without those things, you have nothing better than a divining rod people! Didn't someone call them Saliva Diviners? HT Misha. Oh, and Francis, Told Ya So in 2008!

5 comments:

Keith said...

Dear Steve - agree, genetic exceptionalism is not a good thing, not for disease risk prediction nor for stringent anti genetic testing regulations. The CRP paper was interesting but the same sort of caution is required in over-interpreting what the data means. In the JAMA paper the data do not really argue against a causal link, they simply don't support it, but that is quite a different thing. The could only claim that it argues against a causal link if they also had the bio-data, i.e. the CRP levels as well as the SNPs. I had a brief exchange with a colleague which saves me retyping it all (also the recent paper on "debunking" the SLC6A4 gene variant and depression does no such thing, it doesn't support the hypothesis, but it doesn't prove it wrong either, still work to do on that one):

Interesting – but I’m not sure I agree with the conclusions about CRP not being causal, I don’t think that they can say that based on the studies. The ORs of those SNPs that they do find associated with heart disease are so very small (eg. 1.06 for LEPR) that an OR of 0.98 for CRP I don’t think says much. It’s like with MTHFR – the 677T SNP is often not found to be associated with heart disease or stroke when the studies just look at gene-disease. When they look at folate or homocysteine levels as well as the gene, then they do find the link, but only in those where homocysteine is high. It could be the same with CRP – if you do a massive study but only look at gene and disease then you may or may not find a link – especially for heart disease which is so common. Without looking at the biomarkers or diet (or both) you cannot make strong conclusions – heart disease is so common that a proper control group is hard to find, maybe the controls were generally disease free because they kept the CRP levels down.

In this type of study I think that if they had found an association with CRP SNPs and heart disease then that would argue for a causal relationship. BUT if they don’t find it, as in this case, I think the conclusion is that they cannot say anything – certainly it doesn’t support a causal relationship but at the same time it does not argue against it either

Keith Grimaldi
http://eurogene.biomed.ntua.gr/

Anonymous said...

Did you even read the abstract?

"CONCLUSION: The lack of concordance between the effect on coronary heart disease risk of CRP genotypes and CRP levels argues against a causal association of CRP with coronary heart disease."

In other words, since some SNPs are associated with CRP levels, but not CHD risk, the authors conclude that CRP is associated with CHD risk, but not causally.

So adding CRP level to the Reynolds risk calculator you tout so highly is precisely the same as keeping the sword in a cool place. It is using correlation in prediction of disease risk, without evidence of causation.

Sounds to me like MDs--including yourself--are just as troglodytic as these early adopters.

If you want people to take your "mavericky" attitude and your intent to help people seriously, it would help to make logical arguments to go with it.

Steve Murphy MD said...

@Keith,
I totally agree with you. This study makes this as clear as mud. It is not so much that there is no biologic causation of CRP with heart disease, they argue that there is NONE. I am saying, we are unsure that CRP is a cause. I am cautioning against using NON CLINICALLY validated risk markers for risk prognostication....

So in sum. Totally agree with you. No proof either way, but some proof that to think SNPs which are associated with CRP levels are NOT associated with risk for heart disease

Steve Murphy MD said...

@Anonymous,
Did you even read my blog post?

1. CRP in a CLINICALLY VALIDATED risk model is just fine to use. Unless you know something I don't, the SNPs in the DTC genomic tests have not been clinically validated tools....Damn! Read before you speak!

2. You make my point exactly, there is something wrong with the way we think about risk and CRP and SNPs.......Which you echo by pasting the abstract information and backing it up like a drone bee....

3. I am not saying we need evidence of causation ALL the time, I don't let cats near my kids......BUT, we need to look for clinically validated tools, see point One, thanks

4. I am a little Maverick, big deal. Better to be Mav than off point and shooting completely from the hip without knowing clinical data at all......

Thanks for reading, next time read the whole post and google NeJM Ridker Reynolds Risk

-Steve

Lou Stauffer said...

If genetics play no part in decease please explain the following to me...

My gr-gr-gr grand mother suffered with (quoted from her journal)"dead legs that burned like fire" "fell again today over nothing"

Her oldest daughter, my gr-gr grandmother was said to have the same symptoms only worse. She developed gangrene in her right foot and died @ age 25.

Gr-grandmother, the above only child, lost her left leg after suffering from the same symptoms as those before her.

My grandmother was lucky, she only lost her little toe after having stepped on a thorn and since she had little feeling in her lower extremities since age 20, she overlooked the wound.

My mother, lucked out and seemed to escape this strange unexplained problem.

I was not so lucky. I was diagnosed at 24 via DNA test with CMT Char-cot Marie Tooth, a commonly overlooked genetic peripheral neuropathy. There is no cure and Doctors only treat the symptoms.

I am 51 now and in a wheelchair but I can walk short distances with AFO's and a cane. I had 3 children, apparently symptom free. My only daughter died 3 yrs ago leaving a 15 yr old son.

He was diagnosed yesterday via DNA tests with CMT, the same as my own. He had just accepted a full 4 yr football scholarship to Baylor as a premed.

Genotyping is sweet superstition? I beg to differ.