One thing we learned is that Mr. Young is not so good doing technical cross examination of the other party's witnesses. Press room consensus is that he may have had a few week singles, but not many balls out of the infield.
Except for interruptions, Landis witnesses got to make their scientific case. Unfortunately, we don't yet have Meier-Augenstein's slide set.
Herr Doktor Professor Wolfram Meier-Augenstein was less charitable to the IRMS chromatography than Dr. Catlin was the other day. His key point, on which much of the case may hinge, was that the GCMS and IRMS retentions times weren't in the same county, much less within the mandatory criteria of WADA TD2003IDCR -- retention times matching to 1% or 0.2 minutes, whichever is smaller. The actual variances were on the order of 6% and 6-10 minutes, which raises large questions whether they located the internal references that are supposed to anchor the analysis correctly.
What that means is there is a prima-facie ISL violation, thus flipping the burden of proof onto USADA to show the violations did not cause the reported results.
Now, if things are misidentified, then all bets are off; none of the results is valid.
Granting for sake of argument that the identification of the start is "close enough", and that other peaks are identified more or less correctly, then we get into possible quantification errors.
First we have issues of co-elution, another form of identification problems. Without use of all three parts of the GCMS, 44/45, 2/1 and full-spectrum, we can't be sure peaks aren't co-eluted. If they are co-eluted, we don't know what is going on. Even if we knew what was there, we may not know in what amounts, and while Brenna is working on software to do the deconvolution that would extract numbers, no instruments use it.
(This may be what Catlin was referring to when he observed LNDD was able to do things he couldn't. Herr Doktor was saying they aren't doing it either, because it can't be done, except by assuming there is no interference.)
Second, he claimed that quantifying the internal standard in real samples is important, even LNDD doesn't do it and their procedures don't require it. He claimed it was an incorrerct assumption to believe that the measure in the mix-cal proved the software on the instrument was working correctly when applied to a complex uruine sample. He went round and round with Young, with Young trying to get him to admit the procedure didn't require it, and Herr Doktor resolutely saying they had to, because the didn't have any other way of identfying their internal standard accurately. Eventually Young gave up on that line of questions.
Third, Herr Doktor demonstrated that peaks will have artificially altered CIR values if they overlap, in a way disadvantageous to Landis. When Young tried to impeach that with a Brenna paper, Herr Doktor explained that Brenna's paper was a mathematical model showing possibilities, while the one he cited was real experimental data and confirmed by his experience. He would like to have Brenna's experimental software on his machines, though.
Then, issues of integration points and baseline subtraction come into play. Challenged by Young, Herr Doktor borrowed a calculator and showed that Young's hypothetical would result in a greater than -2 delta unit swing, which was not what Young wanted to hear.
Young gave up, finding cross of Herr Doktor like wrestling a greased pig-- he wasn't getting anywhere, and the Doktor was enjoying it.
After that, we got John Amory, a member of USADA anti-doping review boards, and an expert on testosterone metabolisation. He made the following major points.
- Despite Papp and Catlin's beliefs and suspicions, there's no published papers saying testosterone affects recovery.
- In all of the published studies, 5a and 5b - pdio track each other within 2 delta units, and correlate strongly to t/e changes. Both Young and Suh walked him through charts, and there was exactly one datapoint from one subject in the one paper where this was not so.
- None of Landis' samples showed behavior consistent with the behavior observed above. The 5a and 5b values in the positives were much greater than 2 delta units. In the one case where they were close, the T/E was not elevated at all.
Here's what I think the essence of the case is for USADA:
- One metabolite is good enough, no corroboration is necessary.
- The lab work didn't violate the ISL.
- All these tests show one.
- The lab work obviously violates the ISL for diagnostic ions on the T/E.
- The lab work obviously violates the ISL for peak identification in the IRMS.
- Overlapping peaks can falsely skew the wrong direction on overlapped peaks; Many peaks overlap in the Landis samples.
- Small changes in the integration and background subtraction can have large effects on the reported delta numbers. None of this is recorded, and the results aren't highly repeatable in the reprocessing.
- Unidentified co-elution can have large effects, and there is quite possibly co-elution because LNDD didn't look at everything they could to rule it out.
- The reported numbers are not consistent with what's known about metabolism, in that the 5a and 5b do not track in ways seen in any published data.
Landis is going to offer Simon Davis, an IRMS machine expert, so suggest exactly what is wrong with the way that LNDD does things to produce these odd numbers.
Once the burden is flipped, the question will come down to: Which is more likely, that Landis has this complicated doping regime contradicting known chemistry, or that LNDD is systematically messing up measurements in the way shown by Davis?
With the burden flipped, USADA must prove that LNDD is not the reincarnation of Muppet Laboratories.
The arguments made by Goldberger and Meier-Augenstein may be hard to refute with the traditional, "looks good to me!"