BRUNET: Welcome Back.
CROSS BY YOUNG:
a: none; we don't provide commercial services; we do research only.
q: what research have you done for steroids in urine?
q: no irms reports on steroids in urine?
q: you have various instruments in your lab? don't have an isoprime?
q: familiar with the software of the isoprime?
q: do you look at WADA documents day-to-day?
a: no, doesn't apply.
q: were you paid for your preparing slides?
a: for consulting on the case, yes.
a: 750 pounds/day.
q: did you send drafts to counsel for comments?
q: comments from counsel?
q: anything added?
a: only reference numbers.
q: stuff left out? linearity?
a: left out for time.
q: should do linearity by peak height?
a: no, should use peak area. peak height is not a good representation, it's an approximation.
q: in terms of range of linearity, you should look to be sure the areas...
a: if you are comparing, should be like-for-like, see Dr. Aquilerra's note on this in his experimental section. This is exactly what you should be doing.
q: he wasn't talking about linearity was he...
a: yes, he was
STEPPING ALL OVER EACH OTHER.
SUH: Let the witness finish.
BRUNET: Let him finish asking before answering.
q: was he talking about linearity, or the range of delta values compared to..
[ floyd is enjoying this ]
q: drawn two peaks with good separation [ at -28 ]
q: drawn to that slightly run into each other. You referenced Brenna's paper?
a: yes, used it as an example?
q: Bernna's paper tells us when there is overlap, you'd have this effect?
a: no, I used another paper that showed data; Brenna's paper only for an illustration.
q: Brenna's paper says you'd get something like -29 and -27, correct?
a: mumble, software, deconvolution,
q: I'll show you the Brenna paper, page 1298.
q: figure 5, true figure EE shows substance that comes out first is more negative than the one that comes out second?
[ staring ]
a: this is data from models; the paper I cited is from measured data. At the 95% confidence level, all of those points overlap. My point was that without methods to deconvolute peaks, we don't know.
q: so we don't know which way it will go?
a: in my experience, and in the paper I cite, it's more positive first, more negative later.
q: you are disagreeing with Dr. Brenna?
a: this is a modelling paper, it is not experimental; highly interesting work with algorithms, but nobody does it that way. So what we have as users is what we have. My experience in 16 years in this business, the leading peak is more positive than it should be, and the trainling more negative.
q: when you look at 5aA peak, do you see any interference with that peak? Is this good GC?
a: for GCMS this is acceptable. Small peak between 5a and 5B and something that is a shoulder.
q: small effect -- if this little peak were to be included in the irms measurement of the 5a, that would have a small effect?
a: not clear. if included, the affect on the result is not predictable; weather measureable, depends on the composition of the small peak.
q: is there anything you find in urine that would have an isotope ratio below -40?
a: who is to see it comes from urine? Can you tell me it doesn't come from matrix interference? We don't know, we don't even have the mass spectrum from this peak. Considering your baseline is unbalanced -54, something -129 at LNDD could have effect.
q: you say this peak could have a carbon value of -129?
q: what in nature has that value?
a: it doesn't have to be in nature my dear friend. we could use values down to -700. It's an artifact of the matrix interference, doesn't come out of the urine, we just don't have the information. There's no comment it's a natural peak. No way of knowing.
q: if this was a -70, how much would it change the delta value.
a: you have to know the peak area, make an assumption...
q: is this compared to this 1%, 2%?
a: hard to say, on the tail of the preceding. would not care to guess. we can say it's not 50%. Need the data, impossible to speculate.
q: if it's as high as 5% and -70, it wouldn't change the delta value more than a delta unit?
a: if you give me some data, I'll calculate if for you.
[ digging for calculator .... calculating... ]
a: would affect -3.5
q: you're saying...
q: peak is -31.88, call it -31 or -32.
a: if you assume peak is -28
q: but it's 31..
q: if you're trying to determine contribution, let's find one without overlap...
SUH: he's been given assumptions and a calculator, let him calculate.
q: let's let you run your hypotherical, then we'll let you re-run it.
[ notice can of Red Bull behind Floyd ]
a: -28.75, -70, 5% now is 30.8125, -2/mil change.
SUH: enter that as evidence, please.
q: you talked earlier of CG's you'd find acceptable in your lab?
[ blinks slowly ]
NEW EXHIBIT is 121.
q: Look at figure on page 67 of your paper. used as ex of bad cg?
q: fig 3, ex of good one.
[ zooms in ]
q: tails on these?
q: run into neighbors?
a: those are reference compounds. If you look at the scale you obscured by zooming in, we're at 0.6 na; it's log scale; it's a disproportionate graph.
q: where does the tail end on this peak.
a: doesn't have a tail, it's a reference pulse in the combustion chamber.
q: where does the tail end on this one?
a: somewhere around here. without the scale expansion, you wouldn't see any of these things.
q: what is the time period?
a: not identical for each peak, .... can't really tell you, the figure was for something else, and that's not really there. This is not an application example, could only give you a guess of 40--60 seconds.
q: not a good chromatogram?
a: yes, but not a real sample.
q: when you were talking about the scale, you were saying that blowing them up makes them look worse.
a: it can emphasize the peak tail, that's true.
Q: how many types of GC instruments do you have?
a: hp, tracegas, agilent, varian, 4-5 types of GCs.
q: If I took a internal ref compound like 5aAAC, ran it in two different instruments, would you expect the exact same retention time?
a: if you get same other things, should be within .2 minutes.
q: let me break that apart. same column?
q: same temps?
a: gas flow, column mfgr, ideally same batch.
q: across instruments?
a: that's why you use relative retention times.
q: if running on two instruments, you wouldn't expect retention within 1%?
a: that's why you use .2 minutes.
q: if internal standard in early, you'd expect larger fluctuations?
a: no, shorter fluctuations.
q: standard in early part of injection, you wouldn't be surprised to have differenced > 1%?
a: you don't tie down on one peak, when comparing betwee ninstruments, you look at relative times, to guard against such problems.
q: you look at relative times ...
[ Young draws ]
a: no, we don't.
[ but Young is still drawing. ]
q: when you looked at the results, did you compare the retention and relative for the GCMS instruments, and all consistent?
a: yes, all run on same instrument. If they don't come out together, you'd have a serious problem.
q: that tells you the instrument identifies the IS, it's identified the analyites.
a: on that instrument, yes.
q: so when you look on the IRMS, they were consistent?
q: so the comparison you were making was not across the instruments...
q: instruments accurate within 1%.
q: but acorss is where you got 6%
a: so comparing between instruments is where LNDD had a 6% variation, which is what you shouldn't have. The point of relative retention is to commpare across instruments. The moment you don't, you have a problem. since you are way out of spec, you have lost your anchor on the GCMS.
q: if the purpose of relative times is to compare two instruments, in most analysis using GCMS, why use relative.
a: to get absolute certainty on a run to run basis.
q: when doing simple gcms they use relative retention?
a: they can, also use absolute, also use mass spectra to tell if it goes off. in this instance the purpose is two fold, the GCMS, and require that reference to identify the peak on the IRMS, because you don't have the second level of identification. Need it go QC the instruments and identify, and this has clearly not happened here, it's 6% out.
q: if you were to apply this criteria to the GCMS instrument,
a: including the mass spectrum.
q: did you also talk about CIR of the internal standard, this morning?
a: yes. i did.
q: you know they don't use the IS for quantification.
a: they should use the internal standard in the matrix for QC.
q: if you hear they testify they only use it for identifcation... The delta differences you pointed out as outside their criteria, would only be outside if they used the standard for comparative delta values.
a: the problem is how they have identified the IS, is the time marker, but it's not even close to the time of the GCMS. How on earth did they identify which one out of 5 or 6 is that it. So did they use the delta value to say that's it, but it's outside their limit, on what ground do they say that's the IS?
q: where in their SOP do they say they need to do that?
a: If they don't, I'm baffled by how they identify the peak.
q: where is it in the spec it matters if it's outside the spec?
a: they must have some criteria, how do they choose? Divine intervention? I'm amazed. You're left with the GCMS, which has mass spectra, then you get to retention time, and it doesn't match in the IRMS. How do you identify one unknown peak among 5 unknown peaks? I don't know how they do it.
q: where does it say they quantify the IS?
a: how do they know it is the internal standard?
q: you've already told us your not an expert in steroid metabolism, so we'll skip
[ landis smiles ]
q: page 15 of this study (cologne); you referred to one volunteer for the proposition that there's never a > 2 difference in 5a and 5b. In this volunteer P9, what do you see at measuerment 3?
a: yes, I did but this is not the difference between the 110ha and adio, which is completely irrelevant. you are not comparing like-for-like. All the studies show the same measure. You are comparing apples to pears.
[ Botre sighs. ]
q: so you're point is this is not relevant
a: this is not on any page of the LNDD results.
Q: you have no experience in measuring steroids?
a: but that is not the point, you need to compare corresponding measurements, the measures in Shakleton, Aquilerra dnd cologne are all 5a-pdiio and 5b-pdiol; and this is only one volunteer anyway. I'd like to see any peer-reviewed journal accept a study with n=1. That' you can't compare, I'm terribly sorry.
[ Question in the press room what the over/under is for how many more times Young will make sure we know he isn't an expert on steroids ]
SHOWING chart from reprocessing with the GC and 2/1 trace, but only the 2/1 trace right now.
q: this is baseline drawn by os2 software automatically. is that a good baseline?
a: i have no idea how os2 does this.
q: you don't use 2/1 trace, oh you don't use this software...
a: yes, I do, i use all of them. sometimes the 2/1 tells you sometimes not. It's not foolproof. When you get nonsensical results you try to find out why this is.
q: you said earlier the 2/1 is a good way to identify peak starts and stop, and gave some exceptions.
a: I use the software that lets me define the slope to trigger start stop; you can't with the 2/1 trace, and it's a ratio, which has no units.
YOUNG: you're swinging your pointer around...
a: peak software measures.... on and on. ... how can you possily detect when the start goes up, and the tail also goes up. To say they always peak detect peaks with the 2/1...
[ Brenna rubs eyes; press room hoots ]
a: I don't know how you peak detect using the 2/1 trace.
q: GDC 1313, chart of "strawberry aroma", fig 14.12, where you couldn't use 44/45 to detect.
a: no, this is where the CG shows that you need both 44/45 AND 2/1 gives useful information. So here you still can't detect the peak, but you have an indication there is something probably worth looking at. You still can't find the peak. You'd have to rerun the sample, and normally inject more
BRUNET: corrects YOUNG on document labels, is this one numbered? now GDC 1355(?)
q: not going to take you back to your slide, but the CG here is only the 44 ion. You wouldn't expect to see a peak here, if there was a highly enrighced c13 substance?
a: I don't understand the question. This is an example about peak detection.
q: the reason is this is a 44 ion trace, and what's in the study is found at the bottom of page 44, a highly c13 compound, with virtually no c12.
a: long correction. Read the entire paragraph. There is something here they didn't expect to see that is highly C13 enriched.
NO FURTHER QUESTIONS.
SUH: I'd rather make sure he gets his flight than ask more questions.