Tuesday, February 13, 2007

Is it the contamination?

Discussions all over the place have been trying to make sense of USADA's request to test B samples from A's that were declared negative. We believed (and still mostly do) that it was just a game to waste time and divert resources.

On the other hand, there is the matter of the hidden A-sample results. USADA has been sitting on them for months, claiming they are not relevant. We can't know how relevant they are until they are examined. The concern would be that they contain exculpatory evidence.

What exculpatory evidence could they contain?

The obvious one is longitudinal T/E data, indicating either the S17 results weren't that out of line with negative contemporary results. A more subtle one is that they'd reveal the inaccuracy of the LNDD measurements of epitestosterone in the denominator, which wildly affects the results for low values of E. It's already been pointed out here that the calibration of the E measurement results was assumed to be linear, but is almost certaily curved at the low end of the detection range, resulting in an artificially low E reading that would skew the result.

Today, another thought struck. The A sample results could demonstrate the likelihood of contamination in the S17 test. The values cited by Baker in the slide show to demonstrate contamination are all from the screening assay. If they are radically different from the same values in the other A sample tests, then that would be a strong indication that something was wrong with the S17 sample from the git-go.

Note that we have been consistently skeptical of the contamination argument. This is the sort of data that would change the opinion.

TBV

Update: we just noticed this is our 300th published article. Who would have thunk it?

14 comments:

Anonymous said...

Exculpatory evidence is the evidence favorable to the defendant in a criminal trial, which clears or tends to clear the defendant of guilt. In many countries such as the United States, if the police or prosecutor has found such evidence, he or she must disclose it to the defendant. The prosecution's failure to disclose exculpatory evidence can result in the dismissal of a case.

daniel m (a/k/a Rant) said...

TBV,

This is an interesting speculation about the A sample. I've been wondering the same, myself. And I've been wondering if the A sample, itself, was contaminated (at least, by WADA standards). Haven't had the time to look to see if one can calculate the percentages of free T or free E in the A sample. I've been under the impression that Dr. Baker's analysis has to do with contamination of the B sample. If it's possible to calculate the free amounts in the A sample, and if the results are over WADA's limit, that would cast the case in a whole new light.

- Rant

Anonymous said...

Is there anyone (arbitration lawyers?) or anything that can make USADA hand over the A tests? A good FOI law to be invoked? At the FFF event last week, there seemed to be a hint that this is what Floyd's lawyers were waiting on, and possibly thought they could get eventually. I'm sorry I didn't pay more attention at the time.

DBrower said...

Rant, Arnie's contamination claims are all based on the S17 A samples. The argument is they were obviously contaminated, and should have been thrown out with no testing done at all.

Debbie, it will certainly come up in motions when the arbs start entertaining them. Despite the schedule, it's not clear when they can really start work.

Delaying delivery of the A results is a game of "keep away" by USADA.

TBV

Anonymous said...

I'm sick of the games the USADA is playing, and I want to see some good old US courtroom discovery, etc. The USADA may not be able to play by their own rules when the american people see what happens in the OPEN hearing...and are reminded how much of our tax dollars support this "old west" type of justice.

Anonymous said...

Maybe the USADA feels like they are entitled to collect any data they want provided nobody stops them. That would fit well with their attitude that they don't have to give up any data unless forced into it.

Between the Baker slide show and some of the analysis by guys like Duckstrap and Tom Fine on DP forums, I'm convinced that the IRMS test as applied by LNDD gives an alarming number of false positives. There have to be people at USADA who are aware of this by now, though of course they wouldn't admit that any positive is false. If they test the B samples at LNDD they might get lucky and find another positive. The more samples they test the higher the chance of getting another positive, so why not ask to test a couple of out of competition samples too? It wouldn't be a positive according to the rules, but it would be damaging to Floyd's case anyway.

That's my little fantasy about what might be happening with the request for B-sample testing. It may just be a fishing expedition like a lot of people originally thought.

Anonymous said...

In the Baker slides, the A sample demonstrates the accuracy problems. The B sample demonstrates the evidence of contamination in the form of excess free epitestosterone.

Everyone acknowledges that the screen test for the T/E ratio is less accurate than the confirmation test, but the question would be: If there is sufficient discrepancy between the screen test and the confirmation test, is that an indication of contamination or at least unexpected chemistry in the urine sample, such that the results of any testing would be likely to experience errors of unknown magnitude? What is the expected accuracy of the screen test?

LNDD asserts that the confirmation tests prove that the T/E ratio for the stage 17 sample was greater than 8.0 (USADA0101), assuming a documented potential measured error of -30%. The lowest and first T/E screen test produces a measured result of 4.9 (USADA0054). This requires that the T/E screen test have an expected measured error of up to +63.27%. (And, a threshold of 4.0 for the screen test is actually testing against a threshold of 6.53.) Is this error percentage expected or does it indicate a problem with the sample? Is it reasonable to see complimentary maximum errors in opposite directions for the screen test and the confirmation test?

The lowest measured quantity of E is 5.2 ng/ml (USADA0288) using the confirmation test. A potential error of +30%, as documented, would allow the actual quantity to be up to 6.76 ng/ml. The highest measured quantiy of E is 13.7 (USADA0054) using the screen test. This requires that the T/E screen test have an expected error of up to -50.66% for the quantification of E. Is this reasonable? And once again, is it reasonable that the errors for the two tests would be at their maximums and in opposite directions?

LNDD took the unusual step of re-testing the A-sample with the screen test after all confirmation tests were complete (USADA0057). Presumably, something concerned them about the results of the ealier screen test. What was the concern? (There does appear to be some non-conformance issues, “Concordance des tr”, documented in USADA0068 and USADA0069.) The screen test was not applied to the B-sample. Is it normal for the B-sample to be tested without the screen test?

If the other A-sample tests become available and if anything is to be inferred from the results, the potential error of those results will be need to be discovered. It is likely that LNDD does not want to reveal the expected errors of the screen test.

Anonymous said...

TBV - This jogged my memory on something Floyd posted over at the DP forum. Here is the post - it makes his post much more interesting in light of your thoughts.
http://www.dailypelotonforums.com/main/index.php?s=&showtopic=1752&view=findpost&p=37019

Cal

Anonymous said...

Yes. The last sentence hits this notion on the bullseye.

Anonymous said...

Interesting paper, although a little old (most recent reference is 1999):

http://igitur-archive.library.uu.nl/dissertations/1977561/c1.pdf

"It is suspected that a naturally high TG/EG (T glucuronide/E glucuronide) ratio is associated to a low excretion level of EG compared to a higher excretion level of epitestosterone sulfate (ES) due to stimulated sulfotransferase activity in the testes. As a result, Dehennin [42] therefore suggested the TG/(EG+ES) ratio... It was shown that a physiologically high TG/EG ratio is associated with an EG/ES ratio below 1."

Does the screen test catch ES but the confirmation test ignores ES? This could account for a drastically reduced E measurement in the confirmation test.

Also, a good discussion of contamination.

Anonymous said...

It looks like ES is old news. See:

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=9001956&dopt=Abstract

It looks like LNDD led the way for methanolysis as an alternative that create a combined profile of androgen glucuro- and slfoconjugates.

Anonymous said...

It may be old news, but WADA may only be measuring glucuronides.

See WADA Technical document TD2004EAAS referenced in the wiki:

"The T/E value is given by the peak area or peak height ratio of testosterone and epitestosterone (equivalent to the glucuronide) obtained by measuring the ion at m/z 432..."

Anonymous said...

Is the variation in calibration accuracy a consequence of varying ratios of glucuronide and sulfate conjugates in the control dilutions of T and E? The ratios are not documented. If the calibration sample has a different ratio of conjugates than the test sample, how do you know that the test measurement is accurate?

Anonymous said...

Rant appears to have cited a similar article in October, http://rant-your-head-off.com/WordPress/?p=61 .