Power Point analysis, two critical claims Options

[one-mint-julich]

I have not been able to open the large PDF files, so have only looked at the Power Point presentation, on the TBV site. I want to say, first, that I really respect TBV. He has made no secret of the fact that he is a huge Landis supporter, but his comments on the slides are very objective. He’s doing everyone a great service.

Some of the points made in this presentation I think can be immediately dismissed:

1)The variability in T/E ratios taken from a single individual at different times, said to be up to about 30%, has been discussed before. Floyd’s team provides a slide (16) indicating one sample that was determined at about 4 on one test, then the identical sample determined at 11 later. I don’t know anything else about that sample, but it may be just a case of lab error. The literature shows quite clearly that T/E ratios are far more stable than this.

2) Another point I regard as a red herring is the claim that 3.8 delta units, rather than 3.0, should be used as the criterion for the difference in the IRMS (Slide 20). The criterion is stated in the WADA document as 3.0. Floyd’s team seems to think that since the standard deviation for a series of measurements is around 0.8, the criterion should really by 3.8. This was discussed in this forum before. The 3.0 criterion takes into account variation already. It is supposed to be set high enough so that even accounting for variation, the number is highly significant. In fact, if the SD is 0.8, a value of 3.0 indicates a difference of nearly 4 Sds, which is highly significant.

3) I also would dismiss is that the studies were not blind, because the testers had the TUE data (Slide 23). This was discussed in the forum yesterday. Floyd himself said it was standard procedure to submit TUEs with the A and B samples. In fact, many riders in this year’s Tour had TUEs, so this would not single out Floyd. And even if it did, another poster pointed out that it would be very hard to tamper with the B sample in a way that would not be evident when F;oyd’s rep came to watch it be unsealed.

Having eliminated what I think are attempts to throw a lot of mud and having something stick, I do accept that there are two very critical points made in this presentation.

1) They argue that the ratio of free E to conjugated E was greater than 5%, indicating possible microbial contamination (slides 8-10). They specifically cite a USADA document that says the sample should be disallowed under these conditions.

I haven’t seen this document yet, but the claim is not frivolous. Does a ratio of free to conjugated this high indicate contamination? I have pasted below the abstracts of two relevant articles. The first study found that a ratio of greater than 10%, not 5%, was indicative of contamination, but even then there was no detection of bacterial conversion of conjugated T or E to the free form. The second study found conversion of conjugates to free form when urine (sterile, apparently unlike Floyd’s) was incubated at 37o c for a week, but even then there was no change in T/E.

My opinion: From a scientific point of view, I don’t think the data shown on these slides affects the conclusion. In the first place, I don’t think the 7.7% figure they show indicates contamination. Second, even if it does, and even if the amount of free E all were due to contamination, it is not enough to skew the T/E ratio from negative to positive. Third, if free E is being formed, so most likely is free T, so again, the ratio is probably not much affected.

That is my scientific opinion. It is not substantiated fact. Moreover, legally, Floyd may have a case if there is a hard-and-fast rule about the 5%. This technicality might be sufficient to clear him.

2) The second critical point concerns the four metabolites used in the IRMS test. As I pointed out earlier, the 3.8 o/oo criterion is a red herring, so two of the metabolites, not one as leaked earlier, are over the limit of 3.0. But two of them are below the limit (Slide 21).

How do we rule here? Arguments are focusing on interpretation of the WADA document, whether one or four metabolites are necessary to indicate a positive. I think it’s one, but I think the discussion should go further. Why are there such big differences in these values? Poster Thomas Fine pointed out that in the original Catlin IRMS study, there was a big difference between the isotope ratio of pregnane, a reference metabolite, and the T metabolites, even in controls. The consistently differed by 1.4 – 2.0 delta units. So apparently different metabolites have different ratios, even in individuals who are not taking any synthetic steroids. The developers of the test are presumably aware of this, and this may be why several different substances are listed for making the test, but if this is the case, it has not been explained clearly in any document that I’m aware of. In my opinion, this is a serious shortcoming of WADA (unless it is explained in some doc I’m not aware of).

The data for Floyd shown in slide 21 may be typical of a testosterone user. Maybe it’s normal for some comparisons to be over three, and some not. But this has to be clarified by the developers of the test, and specified in the WADA documents. In particular, the kind of “bias” seen when using pregnane needs to be addressed. Is this the case with other reference compounds as well? Can it go in either direction? Are some reference compounds likely to lead to large differences in normal individuals, whereas others go the opposite way, resulting in small differences in testosterone users? There’s nothing wrong with this if the differences are well-known, consistent, and clearly stated.

My opinion: Though two of the four comparisons are not over the criterion, they are both over a value of two delta units, which probably indicates about 2.5 Sds. This is highly significant, and together with the two values that are well over the criterion, are highly suspicious to me. The one value of greater than six I find it hard to believe would ever be found in someone not taking a synthetic substance. It is well beyond explanation by the pregnane bias. That seems quite damning to me.

But I’m still troubled by the pregnane bias, which could account for most of the difference in the other comparison using pregnance, and which opens the door to questions about the other two comparisons using a different reference (not a cortisol metabolite, by the way). I think it is really important to have control data from Floyd at other times. Just seeing the problems here, I think UCI ought to consider getting these data on a regular basis from riders. I know, it’s very expensive to do this, but a set of controls could probably erase all doubts one way or another.

Beyond the scientific opinion, the technical question revolves around one vs. four comparisons. But I hope that the case does not devolve to this, because that will obscure the questions about the large differences between the different comparisons.

References on contamination of urine

Anal Biochem. 2001 Feb 15;289(2):116-23. Related Articles, Links

Changes in androgenic steroid profile due to urine contamination by microorganisms: a prospective study in the context of doping control.
de la Torre R, de la Torre X, Alia C, Segura J, Baro T, Torres-Rodriguez JM.
Pharmacology Research Unit, Institut Municipal d'Investigacio Medica (IMIM), Doctor Aiguader 80, E-08003 Barcelona, Spain. rtorre@imim.es
Urine contamination by microorganisms may affect the interpretation of urinalysis in different areas of clinical diagnosis. This is particularly relevant in doping control. A prospective study was designed to assess the effects of urine contamination by selected pathogens on the endogenous androgenic steroid profile. Pooled urine from a healthy male volunteer with standard steroid profile compared with reference values for the Caucasian population was sterilized by filtration and stored in sterile glass tubes. Aliquots were inoculated with known amounts of 15 different organisms (bacteria, fungi, and moulds) and incubated at 37 degrees C for 2 weeks. Different markers of urine contamination, such as pH, deconjugation of steroids, and metabolic by-products, were determined. Alkalization of urinary pH was not a reliable indicator of urine contamination as several organisms grew in this medium and no alteration of this parameter was found. In uncontaminated urine, less than 10% of steroid glucuronide conjugates were spontaneously hydrolyzed. Higher rates of hydrolysis for sulfate conjugates were found. An unconjugated fraction higher than 10% of the total amount of testosterone was a reliable indicator of urine contamination. However, microbial production of testosterone or epitestosterone was not detected. In contrast, a few organisms were able to synthesize 5alpha-androstanedione, 5beta-androstanedione, and androstenedione using endogenous steroids as substrates.

Stability studies of testosterone and epitestosterone glucuronides in urine.
Jimenez C, de la Torre R, Segura J, Ventura R.

Unitat de Farmacologia, Institut Municipal d'Investigacio Medica, Barcelona, Spain.

The stability of testosterone glucuronide (TG), epitestosterone glucuronide (EG) and the T/E ratio in urine has been studied. Samples were analyzed by gas chromatography coupled to mass spectrometry (GC/MS). Urine samples were submitted to a solid-liquid cleanup followed by extraction of unconjugated testosterone (T) and epitestosterone (E) with tert-butyl methyl ether (free fraction). The remaining aqueous phase was hydrolyzed with beta-glucuronidase and extracted at alkaline pH with n-pentane. Analytes were analyzed by GC/MS as their enol-trimethylsilyl (TMS) derivatives. The urine for stability testing was obtained from an excretion study after the administration of T to healthy volunteers. The homogeneity of the sample was verified before starting the stability study. The stability of TG and EG was evaluated at different storage conditions. For long-term stability testing, analyte concentration in urine stored at 4 degrees C and -20 degrees C was determined at different time intervals for 22 months. For short-term stability testing, analyte concentration was evaluated in urine stored at 37 degrees C for 3 and 7 days. The effect of repeated freezing (at -20 degrees C) and thawing (at room temperature) was studied for up to three cycles. Data obtained in this work demonstrated the stability of TG, EG and the T/E ratio in sterilized urine samples stored at 4 and -20 degrees C for 22 months and after going through repeated freeze/thaw cycles. Decreases in concentration were observed after 7 days of storage at 37 degrees C due to the partial cleavage of the glucuronide conjugates; however, the T/E ratio was not affected. These results show the feasibility of preparing reference materials containing TG and EG to be used for quality control purposes.

[shag]

1) They argue that the ratio of free E to conjugated E was greater than 5%, indicating possible microbial contamination (slides 8-10). They specifically cite a USADA document that says the sample should be disallowed under these conditions.

I haven’t seen this document yet, but the claim is not frivolous. Does a ratio of free to conjugated this high indicate contamination? I have pasted below the abstracts of two relevant articles. The first study found that a ratio of greater than 10%, not 5%, was indicative of contamination, but even then there was no detection of bacterial conversion of conjugated T or E to the free form. The second study found conversion of conjugates to free form when urine (sterile, apparently unlike Floyd’s) was incubated at 37o c for a week, but even then there was no change in T/E.

My opinion: From a scientific point of view, I don’t think the data shown on these slides affects the conclusion. In the first place, I don’t think the 7.7% figure they show indicates contamination. Second, even if it does, and even if the amount of free E all were due to contamination, it is not enough to skew the T/E ratio from negative to positive. Third, if free E is being formed, so most likely is free T, so again, the ratio is probably not much affected.

That is my scientific opinion. It is not substantiated fact. Moreover, legally, Floyd may have a case if there is a hard-and-fast rule about the 5%. This technicality might be sufficient to clear him.

From a legal (not scientific) perspective, assuming there is indeed a rule stating that if the ratio of free E to conjugated E is greater than 5% then the sample in question should be deemed contaminated and therefore disallowed, and assuming that the 7.7% figure has been correctly determined, I don' t see how the case goes forward. Whether there might be scientific arguments to be made that it should be 10%, not 5%, or that such contamination would not likely affect the T/E, is wholly superflous. If the rule fits, you must acquit!

If the rule fits, you must acquit!

*No comparison between floyd and OJ is intended.

This post has been edited by shag: Oct 12 2006, 03:57 PM

[Steve in ATL]

2) Another point I regard as a red herring is the claim that 3.8 delta units, rather than 3.0, should be used as the criterion for the difference in the IRMS (Slide 20). The criterion is stated in the WADA document as 3.0. Floyd’s team seems to think that since the standard deviation for a series of measurements is around 0.8, the criterion should really by 3.8. This was discussed in this forum before. The 3.0 criterion takes into account variation already. It is supposed to be set high enough so that even accounting for variation, the number is highly significant. In fact, if the SD is 0.8, a value of 3.0 indicates a difference of nearly 4 Sds, which is highly significant.

The +/- 0.8 is not a standard deviation, which in statistical terms implies a very specific meaning. The +/- 0.8 is a tolerance, meaning a range of error that is possible or permittable by the machine perfroming the measure.

[Will]

<The MOD comes in and snips to save vertical scrolling>

Sounds like you are coming to conclusions before you've gone through it.

Tsk, tsk. We said don't do that. Now you are acting like you work at USADA!

Get the PDF's and work them over. It is all there. Get the WADA docs. Work them over.



MOD Note: In the future, please quote only what you are going to directly address - this saves vertical space! This is directly addressed in the posting ettiquette section. Thanks! The Mods...

[Maddog2]

From a legal (not scientific) perspective, assuming there is indeed a rule stating that if the ratio of free E to conjugated E is greater than 5% then the sample in question should be deemed contaminated and therefore disallowed, and assuming that the 7.7% figure has been correctly determined, I don' t see how the case goes forward. Whether there might be scientific arguments to be made that it should be 10%, not 5%, or that such contamination would not likely affect the T/E, is wholly superflous. If the rule fits, you must acquit!
*No comparison between floyd and OJ is intended.

but it is harmfull to drink 10 liters of OJ.

Back on topic- I'm a little bothered that the presentation is based somewhat on what they assume is the criteria for a positive test- as pointed out above by OMJ- example being does 1 or 4 metabolites needed for a positive...
Assumptions are never good in science

A question to FLoyd- does your team have access to WADA/UCI criteria for the test positivity, and do you have any experts in steroid testing on your team other than our crack team of experts.

I was always struck by Hamiltons case in that he did not have a Hematologist/flow cytometric expert on his side. A respected researcher but not an expert in the area. On the other hand, maybe he didn't have much to work with....

[Bill Hue]

3) I also would dismiss is that the studies were not blind, because the testers had the TUE data (Slide 23). This was discussed in the forum yesterday. Floyd himself said it was standard procedure to submit TUEs with the A and B samples. In fact, many riders in this year’s Tour had TUEs, so this would not single out Floyd. And even if it did, another poster pointed out that it would be very hard to tamper with the B sample in a way that would not be evident when F;oyd’s rep came to watch it be unsealed.


Don't discard this so quickly. The issue isn't whether they roostered up the sample but whether knowledge of identity enables them to cherrypick results or required them to not use the cortisone metabolite. This goes to timing and knowledge, motive and opportunity and does or should (this is UCI/WADA and they are NOT big on procedural protection) make a really big difference.

[jr.]

the WADA rules and recommendations on test reporting of t/e are pinned up top

relevant portion re contamination (see page 2 of the technical document):

The urine Sample is not collected under sterile conditions, and where the circumstances are
favourable, the microbes present in the Sample can cause changes to the profile of the urinary
steroids. Initially there is cleavage of the glucuronides and sulfates followed by modifications of
the steroids’ structure by oxido-reductive reactions. To report an Adverse Analytical Finding of an
elevated T/E value, testosterone or epitestosterone concentration or any other endogenous steroid
parameters, the concentration of free testosterone and/or epitestosterone in the specimen is not to
exceed 5% of the respective glucuroconjugates. Elevated amounts of 5a- and 5ß-androstan-3,17-
dione in the free form also indicate microbial degradation.



Presuming the power point page to be the accurate measure of t and e to their glucoroconjugates, ie, over 5%, then according to my reading of this rule, the initial test could not be determined to be an Adverse Analytical Finding, game over, no IRMS.

In addition, the three tests for that day are Pereiro, Floyd and an unknown random rider. In each stage, since if I understand correctly that dates of samples are provided, two out of three people are known, just not matched to their samples. Now if Pereiro and Floyd have identical TUE's, then that makes things truly random. But I think the testing dates on some of those pages are the 21st and maybe 25th (not going back to look now), so maybe they have a new set of samples they know include Floyd's sample from the tt and TUE and maybe now its not so random. So they are reasonalby certain they have the probable winner of the Tour with a contaminated sample which should not be the basis of an adverse analytical finding. But if you are the lab and convinced there is reason to believe that the winner is doping, do you just sit on it or do you tell WADA and ASO and UCI? Then does the political pressures of people with a mission, maybe all well intentioned, who think they have a cheat, but are stymied, cause them to decide the ends of exposing the cheat justify the means. After all Ressiot lied to get Armstrong's nos. and nobody really blames him it was in a good cause, right? Whistleblowers frequently break employment agreements (government employees frequently leak classified materials), but its in a good cause and nobody blames them (think Deep Throat).

[Bill Hue]

JR-Better and more concise analysis than mine-good stuff.
Bill

[CapeRoadie]

...but it is harmfull to drink 10 liters of OJ...

Well, really that depends on when you are drinking it, and over what period of time, doesn't it, Ferrari, er um, I mean, Maddog?

Nice post, one-mint. I agree about what you stated here:

"1) They argue that the ratio of free E to conjugated E was greater than 5%, indicating possible microbial contamination (slides 8-10). They specifically cite a USADA document that says the sample should be disallowed under these conditions.

"I haven’t seen this document yet, but the claim is not frivolous. Does a ratio of free to conjugated this high indicate contamination? I have pasted below the abstracts of two relevant articles. The first study found that a ratio of greater than 10%, not 5%, was indicative of contamination, but even then there was no detection of bacterial conversion of conjugated T or E to the free form. The second study found conversion of conjugates to free form when urine (sterile, apparently unlike Floyd’s) was incubated at 37o c for a week, but even then there was no change in T/E.

"My opinion: From a scientific point of view, I don’t think the data shown on these slides affects the conclusion. In the first place, I don’t think the 7.7% figure they show indicates contamination. Second, even if it does, and even if the amount of free E all were due to contamination, it is not enough to skew the T/E ratio from negative to positive. Third, if free E is being formed, so most likely is free T, so again, the ratio is probably not much affected.

"That is my scientific opinion. It is not substantiated fact. Moreover, legally, Floyd may have a case if there is a hard-and-fast rule about the 5%. This technicality might be sufficient to clear him".

We'll call it the 5% rule.

[one-mint-julich]

Someone sent me a PM saying I should open the individual PDF files. But I can't open those, either. Anyone else having problems with them?

The +/- 0.8 is not a standard deviation, which in statistical terms implies a very specific meaning. The +/- 0.8 is a tolerance, meaning a range of error that is possible or permittable by the machine perfroming the measure.

Yes, but the same reasoning applies. The 3.0 standard is supposed to take into account tolerance limits. And in fact, tolerance is operationally very close to standard deviation.

I might add that if there were bacterial contamination, it could possibly affect IRMS, since these bugs can synthesize other metabolites, and who knows what kind of isotope ratios they have? Most likely would not impact much on the values shown, though.

The argument about TUEs. Do the testers in the lab know from a particular sample's number that it came from a particular stage on the Tour, or even from the Tour, for that matter? I continue to think that these people who are convinced that someone was trying to frame Floyd are barking up the wrong tree (or roostering up the wrong henhouse).

[floyd]

Someone sent me a PM saying I should open the individual PDF files. But I can't open those, either. Anyone else having problems with them?
Yes, but the same reasoning applies. The 3.0 standard is supposed to take into account tolerance limits. And in fact, tolerance is operationally very close to standard deviation.

I might add that if there were bacterial contamination, it could possibly affect IRMS, since these bugs can synthesize other metabolites, and who knows what kind of isotope ratios they have? Most likely would not impact much on the values shown, though.

The argument about TUEs. Do the testers in the lab know from a particular sample's number that it came from a particular stage on the Tour, or even from the Tour, for that matter? I continue to think that these people who are convinced that someone was trying to frame Floyd are barking up the wrong tree (or roostering up the wrong henhouse).

They know the race, the stage, and on what day and in what dose the sample # recieved any particular TUE.

[rational head]

They know the race, the stage, and on what day and in what dose the sample # recieved any particular TUE.

Vrijman in his report argued that the coding system on the samples can be guessed by an experienced
lab person who knows what he is looking at. He explained more. I tend to think that's true. But the question of tempering is still hard to fathom...

[Steve in ATL]

Someone sent me a PM saying I should open the individual PDF files. But I can't open those, either. Anyone else having problems with them?

Try this page. Let me know if it works for you. Thanks to db...

[Lonnie]

They know the race, the stage, and on what day and in what dose the sample # recieved any particular TUE.

Personally, as a scientist, that bothers me just a little bit. Know the testing regimes at the end of a stage, it doesn't take much of a leap to start speculating on to whom the sample belongs. For a truly random sample, they should have duplicates, spikes and blanks all in the same batch with the distribution not know to the technician running the test. The associated sample numbers should be know only to a separate technician. If you want true transparency, one tech sets up the tests, the other runs it and reads the results....


L

[Diva3]

They know the race, the stage, and on what day and in what dose the sample # recieved any particular TUE.

Geez, why not just slap your name on it?

[shag]

For a truly random sample, they should have duplicates, spikes and blanks all in the same batch with the distribution not know to the technician running the test. The associated sample numbers should be know only to a separate technician. If you want true transparency, one tech sets up the tests, the other runs it and reads the results....
L

Why would they want to go to all of that trouble? It's only the riders' careers and reputations we are talking about, after all.

[jr.]

Not necessarily tampering, RH, but deciding to proceed through testing instead of discarding their efforts because of contamination at a level beyond the rule. Maybe they even agree with OMJ, 5% is too narrow, they are comfortable scientifically its not enough contamination to affect the test results. Of course, I don't know enough science, and haven't read all the pages of the whole report. Just the pdf of that one page. But if it is the final result of the test for contamination and accurate, its very damning for the lab, UCI, etc. Why is this case going forward. So is there a scientific reason or for someone who has managed to down load and read more, a different page that explains or eliminates this issue?

[one-mint-julich]

I’ve read the PDF (ADRB) on the IRMS. This document lists all the isotope ratios for four T metabolites and two controls, in the A and B samples. They list both “true” and “corrected” scores. They don’t explain the difference, but the corrected scores may take into account changes due to derivitization of the substances. It doesn’t much matter, because when they then calculate the differences between T and reference metabolites, they come out about the same (E, A, 5a and 5b are all T metabolites, while 11K and P are reference metabolites):

A (true) A (corr.) B (true) B (corr.)

E – 11K -2.33 -2.57 -1.83 -2.02
A – 11K -3.61 -3.99 -3.18 -3.51
5b – P -2.21 --2.15 -2.63 -2.64
5a – P -5.51 -6.14 -5.72 -6.38

They then claim three flaws:

1) only one of the four metabolites are over the limit of 3.0. I discussed this earlier, including the claim that the A – 11K values not be considered over 3.0, because of the tolerance limits. They go into great detail to back their claim that four metabolites are supposed to be over the limit. I find the arguments typically lawyerly and for the mot part unconvincing (e.g., using the phrase “metabolite(s)” to indicate plural testing is required). However, others may disagree.

2) The 5b – P value should be greater than the 5a – P value, when in fact it isn’t. Their claim is based on the original Catlin study, in which it was the case that 5b – P was higher than 5a – P. However, that study itself cited another study in which the opposite was found. Moreover, they misquote Catlin, saying he said that 5b – P should be higher or that it should be used. What he actually said was that 5b – P was a more robust measurement, because in his study it was higher. He did not say that it should always be preferred to 5a – P, he preferred it because in his hands, 5b was greater than 5a. In fact, Catlin didn’t find a very large difference between 5a and 5b, and clearly there were substantial individual differences, as opposed to the statistical trend he observed. So I don’t find this claim very strong.

3) Their third claim, if I understand it correctly, has real teeth to it. They claim that Floyd had negative controls done in which his 5a levels were measured. They don’t say where they got these samples from, but refer to another document, which may be one of ones I still haven’t been able to open. If anyone here has information on that, including how they know this was a truly negative sample, I would like to know.

In any case, the 5 a values determined in this urine sample were –28.3 and –28.4 (A and , and they correctly note that this is at the edge of the lowest values obtained in any study of control subjects. This compares, for example, to the mean value of 5a in Catlin’s study, from 73 subjects, of –26.35, with a lowest value of I think –27.9. From this they conclude that there was something wrong with the measurements of 5a—that it measured too low--which in turn negates all the 5a – P values.

I think this is a serious charge, but its hard to evaluate for a couple of reasons. First, how can one be sure the urine supplied at the time was when Floyd was clean? Were there independent indicators of that? Second, what kind of error would result in such large wrong measurements for one substance, but not another?, Third, even if these were his negative values, his “positive values” (true, not corrected) for 5aA were around –32. These are still considerably lower than the negative values. So if the negative values really are from urine that can presumed to be negative, his “positive” values indicate a very large change of about –4 delta units. This is damning in itself, even though longitudinal comparisons like this are not allowed in assessing guilt.

Finally, even if you disallow 5a, you still have the A-11K score, which is over (more negative than) –3.0. As I noted earlier, they only exclude this by taking into account the tolerance, which I don’t think is allowable. ,Moreover, the other two comparisons, which are in the –2 – 2.5 range. In fact, the raw scores they provide for all four metabolitesare all lower, without exception (A and B, true and corrected) than the scores for either of the reference compounds. I think a mega-analysis (not allowed) would suggest high significance.

So overall, reading this PDF does not change the conclusions that I drew earlier. The variability of isotope ratios among different compounds needs to be addressed. Scientifically, I think this case is near the border, but legally--here, as well as in the 7.7% free E--Floyd might have a very strong case.

Addendum: What am I thinking? Floyd is here, I hope. Floyd, tell me where this "negative" sample came from? Was it from another stage in the Tour? Was it negative for T/E?

This post has been edited by one-mint-julich: Oct 12 2006, 05:59 PM

[Thomas A. Fine]

1)The variability in T/E ratios taken from a single individual at different times, said to be up to about 30%, has been discussed before. Floyd’s team provides a slide (16) indicating one sample that was determined at about 4 on one test, then the identical sample determined at 11 later. I don’t know anything else about that sample, but it may be just a case of lab error. The literature shows quite clearly that T/E ratios are far more stable than this.

Did I misread that? I thought they got the two different values from A and B tests of the same sample. Which is a horse of a different color.

2) Another point I regard as a red herring is the claim that 3.8 delta units, rather than 3.0, should be used as the criterion for the difference in the IRMS (Slide 20). The criterion is stated in the WADA document as 3.0. Floyd’s team seems to think that since the standard deviation for a series of measurements is around 0.8, the criterion should really by 3.8. This was discussed in this forum before. The 3.0 criterion takes into account variation already. It is supposed to be set high enough so that even accounting for variation, the number is highly significant. In fact, if the SD is 0.8, a value of 3.0 indicates a difference of nearly 4 Sds, which is highly significant.

I know this was corrected as error bars. And you said you still weren't convinced. Think it through. My andro comes out at -28 and my preganediol at -24, both with +/- 0.8. So the -28 could be -27.2 to -28.8, and the -24 could be from -23.2 to -24.8. So the actual range of differences here is anywhere from 2.4 to 5.6 per mil. This is an absolutely huge error for doing such a calculation. Another way to say this is that subtracting one value from the other doubles the error bar. The error on the difference is therefore 1.6. Which based on the various things I've seen corresponds to about two standard deviations. In what universe could a 3 per mil trigger be wide enough to protect against an error that large? Not ours.

And don't try to claim that if one sample is off in one direction, the other probably is too. Error bars are explicitly not supposed to cover sytematic errors, only measurement errors.

3) I also would dismiss is that the studies were not blind, because the testers had the TUE data (Slide 23). This was discussed in the forum yesterday. Floyd himself said it was standard procedure to submit TUEs with the A and B samples. In fact, many riders in this year’s Tour had TUEs, so this would not single out Floyd. And even if it did, another poster pointed out that it would be very hard to tamper with the B sample in a way that would not be evident when F;oyd’s rep came to watch it be unsealed.

The head of the lab, to the worker bees when testing the B sample:
"This is the biggest, most important test you will ever do in your entire career. So make absolutely sure that you don't think about who's sample this might be, and whatever you do don't compare to the original test values. Because after all, if there's any problem here, it could mean all of our jobs."

Yeah one mint, I see what you mean. I wouldn't worry about the lack of blindness either.

My opinion: Though two of the four comparisons are not over the criterion, they are both over a value of two delta units, which probably indicates about 2.5 Sds. This is highly significant, and together with the two values that are well over the criterion, are highly suspicious to me. The one value of greater than six I find it hard to believe would ever be found in someone not taking a synthetic substance. It is well beyond explanation by the pregnane bias. That seems quite damning to me.

I think that the 3 per mil standard is very poorly set in general. On top of that, I think the +/- 0.8 error bar is huge, more so than people seem to be dealing with. And I think that the utter lack of consistency between different metabolites is very troubling.

But I’m still troubled by the pregnane bias, which could account for most of the difference in the other comparison using pregnance, and which opens the door to questions about the other two comparisons using a different reference (not a cortisol metabolite, by the way).

I think you're wrong. 11-ketoetiocholanolone IS a cortisol metabolite, according to my references.

tom

Addendum: What am I thinking? Floyd is here, I hope. Floyd, tell me where this "negative" sample came from? Was it from another stage in the Tour? Was it negative for T/E?

The way I read that, I got the impression that the negative control is from an anonymous non-racing person known to not be using. In other words, someone like the lab tester or Dick Pound pees in a bottle for a general comparsion. Maybe it's even an average of a bunch of people's piss.

If this is the case it would be EXTREMELY damning (against the lab). It would mean that the negative control tested positive for doping on the same test compound that allegedly nails Floyd.

Does anyone know where the negative control comes from? Floyd?

tom

[sundaymorning]

1)The variability in T/E ratios taken from a single individual at different times, said to be up to about 30%, has been discussed before. Floyd’s team provides a slide (16) indicating one sample that was determined at about 4 on one test, then the identical sample determined at 11 later. I don’t know anything else about that sample, but it may be just a case of lab error. The literature shows quite clearly that T/E ratios are far more stable than this.

Did I misread that? I thought they got the two different values from A and B tests of the same sample. Which is a horse of a different color.

from looking at the raw data (usada_b through usada_d) it seemed to me that the ratios were 50/11 (4.5) and 61/14 (4.4) for the A sample (run in duplicate) and 65/4 (16.8) for the B. the B was run on a different machine. i may very well be confused.