The science of the EPO test

garnetstar

Hi all,

I've never posted here before, so I haven't read all the arguments. And, I don't follow cycling; I've never even been too aware of who Lance Armstrong was, certainly couldn't have named any other pro rider.


But, I'm a scientific researcher. When I saw that some new test for a protein, EPO, was causing a lot of news, I checked it out. So, here's my scientific opinion of this test.


I downloaded the paper from the scientific journal Nature (NATURE 405 (6787): 635-635 JUN 8 2000) in which the authors described the new test and the results that they got from testing samples from the 1998 Tour. It was immediately obvious why they're looking to improve this test: it's got a big hole in it.


Essentially, the new test still relies on subjective judgment and interpretation of the results to decide if something is "positive" or "negative" for artificial EPO. It isn't like the EPO is either there or not there. Nearly all the samples give ambiguous, overlapping results; the scientists must then interpret what the data mean and if they lie towards the positive or negative side.


Of the 102 urine samples from the 1998 tour that the sceintists had, 28 were said to have some indications of EPO by the older, more unrealiable test.

Of those 28, they tested 14 that seemed to have higher EPO concentrations, so that they could get less ambiguous data.


Of those 14, only 2 were published in the Nature paper, both of which *were* unambiguous. That's certainly because the paper's reviewers objected to conclusions of positive or negative based on the subjective interpretation of the other 12 ambiguous tests.


If I'd been reviewing the paper, I also would have said that the ambiguous ones could not be published, nor could any claim of positive or negative findings be made about them.


So, it seems that of the 14 urine samples that had, by the older testing method, the most EPO, only 2 gave unambiguous results. Not great.


This test is not ready for screening large numbers of samples; of the 102 samples, only 14 were tested, and only 2, apparently, gave clear positive or negative answers.




Subjectively-interpreted data is usually rejected because it is *notoriously* faulty, and nearly always skewed by the researcher's own ideas.


The way these scientists are looking to make their test better is by finding a better method that separates artificial EPO signals clearly and definitively from natural EPO signals, so that no subjective interpretation of the data will be needed.



So, until the testing from the 1999 Tour is reviewed and published in a scientific journal, and we can all examine it, the alleged results from that study are invalid; nothing at all has been proven, and, scientifically, these results don't even exist.


Without outside reviewers examining the scientists' claim of positive and negative tests, no one will be able to say which samples can rightly be interpreted as positive, and which as negative, and which are just too ambiguous to be reliably judged.


If I was reviewing the paper reporting the results from the 1999 Tour testing, I'd require that all the ambiguous results not be published, as I say. But, I'd also send the paper back and not allow publication until the authors had controlled for the effects of exposure of the urine donors to high doses of platinum or other heavy metals, in their tests. High-dose heavy metal exposure should affect the results of this test, possibly even years after the last exposure.


Just as an aside--urine doesn't have much DNA in it (I believe); I think that it is a poor candidate for acquiring enough DNA to do a test (not certain about this, though). EPO has no DNA; a DNA sample acquired from urine would show who donated that urine (if the test's possible, as I say), but would never distinguish being artificial and natural EPO.



Now, as I say, I have no opinion of Lance Armstrong or the state of doping in cycling, or whether there's lots of circumstantial evidence of doping, or whether cyclists dope, or need to dope, or anything. I don't know anything about any of that. I'm just giving you my scientific analysis of this particular testing method.


That we'll ever see the 1999 Tour data properly reviewed and published is unlikely. Appearance of scientific results in a newspaper, or in any public source, before submission to a journal, is considered a prior publication of data, and most journals will not accept papers whose data has been previously published. Nature sure won't.


Also, the lab claims that its data was leaked to newspapers, but no journal will accept a paper submitted by a lab that has been careless enough to have leaks occur. The lab's scientific credibility, about that data, will be regarded as zero.


Well, JMHO. Thanks.

garnetstar

I just found this New York Times article which talks about the same things as I do in my post above:


http://www.nytimes.com/2005/09/05/s...1&ex=1130472000


It's written in English, not science-speak like my post is, so might be more understandable.

The article is still free to read (with registration), but probably not for long.

mymymy

Garnetstar,
How would you rate an antibody test for EPO in a horse? I understand that in race horses they use an antibody test against human albumin in the equine.
Mymymy