SS - The problem was though that the method by which they caught him was decidedly dodgy.
How would you feel about police breaking a couple of bones to get a confession out of a suspect?
I don't think that's a fair analogy. More like the found a suspect with the stolen goods, but they messed up the search of the property, and didn't follow procedure. The fact was there was synthetic testosterone in his system.
Was there? There were a series of GC-MS spectra which the lab claimed to interpret to show a single metabolite of testosterone with an isotope ratio inconsistent with control compounds. The other three metabolites examined did not show this anomaly. The methodology the lab used to quantify the isotope ratios was not documented (ie no specific procedure). Instead they just took what 'looked right'.
There were issues with whether this was appropriate: whether the peaks identified as the metabolite had been properly identified, and whether they had been properly quantified, a difficult process when peaks overlap. These are the deviations from protocol which the lab have to demonstrate did not adversely impact upon the finding. They did not do so and the judgements illustrate this.
Sorry to delve into the technical, but there is no simple test to say whether there was synthetic testosterone. Instead it is based on inference.
Background:
Endogenous (human produced) testosterone is produced dynamically over a rapid timescale. The ratio of carbon12 to carbon13 in these peaks reflects that of the surrounding environment.
Synthetic testosterone is produced from plant material. This is biased in composition as it has been around a lot longer. So the testosterone it is alleged Landis used would have been marginally heavier - the carbon atom distribution would be slightly shifted.
The test does not examine testosterone. Instead it looks at a number of common breakdown products which are excreted in the urine. These can be separated on a gas chromatograph - feed a tiny bit of the sample in one end and different compounds will take different times to come out of the other end whereupon they are fed into a mass spectrometer. It has to be established that the retention time of the product analysed is sufficient to accurately identify the metabolite. Different metabolites may be isobaric (have the same mass) but have different retention times due to the chemical arrangement of the atoms in the metabolite giving them different properties.
In the mass spectrometer the distribution of the carbon isotopes can be measured by examining the peak area for the ion current in the mass to charge ratio. There will be a big peak, corresponding to predominantly carbon 12, and subsequent peaks corresponding to one atom being C13, two atoms being C13 etc. As the isotope ratio would be about 1% you can establish the approximate peak areas.
The difference between synthetic and endogenous testosterone is based on the relative sizes of these peaks. Any error in measuring the area will give an error in the final result. The machines have a non-negligible non-constant background signal that must be corrected for. It is obvious that the same error in measurement of the small peaks corresponding to the heavy atoms will have a significantly greater impact than any error in measuring the big peak. The difference one expects to see is very small. And it would be diluted by any endogenous testosterone. He did not have an elevated testosterone level (which one would expect to see with this) and the IRMS test should not be used independently, but to confirm that elevated testosterone derives from an exogenous source.
There are algorithms for automatically subtracting background, and for separating overlapping peaks. The LNDD techs used one called 'experience and judgement' (draw the line where they think it looks best) which is undocumented and inappropriate for this assessment as it leads to inconsistencies between runs. Conveniently the original trace data was destroyed/not retained so independent analysis is not possible. Neither was any assessment of error performed.
An adverse isotope ratio was observed in one metabolite, but not in the other three examined. Landis case is that this finding was not justified as the procedure was inappropriate to support those conclusions. Also that there were inconsistencies between the four metabolites from the same sample so the test was unreliable.
I don't know if he doped or not, but the testing as performed does not support the allegations because it was done badly.
Well done if you have managed to get this far.
..d