Monday, August 04, 2008

Irregular Report 4

Blogs
Rant notes the "difficulties" of having real runners in the family, and then cites the recent revelations of the old "switch the urine" trick employed unsuccessfully by several Russian Olympians. Seems DNA now renders this subterfuge virtually impossible. Included is a quote from anti-doping expert Dr. Gary Wadler.

Racejunkie passes along some "hints from Heloise" er um Ricardo Ricco in the form of "advice" to the peloton learned from his recent "indiscretion" Here are a couple of samples:

1. Taking EPO for an entire Tour is a "youthful mistake": Sort of like being an hour late for curfew one night and briefly making your parents worry you've been in a car accident, except with precisely-calibrated IV drips, clandestine trips to prestigious university medical clinics, huge payoffs through dummy bank accounts, and months of strategic planning.

2. You might want to tweak the testing protocols, because I should really have tested positive *every* day: !@#$ if I'm gonna be the only sap in the peloton who goes down for this!

In case anyone is even remotely interested in keeping track of our new cats here is the blog I plan to try to update daily to chart their progress and adjustment.

19 comments:

("Eightzero") said...

http://mine.icanhascheezburger.com/view.aspx?ciid=1702314

Mike Solberg said...

For anyone who can still stand this level of detail:

One of the big issues in the Landis case from way back in the fall of 2006 was the disturbing fact that the original Lab Documentation Packet did not contain an important set of data needed to support the veracity of the Adverse Analytical Finding. A necesssary (but not sufficient) piece of information was the "complete mass spec. data" from the GC/MS part of the IRMS test. This data is necessary to show that the peaks that were being measured in the IRMS test were "pure," that is, that they contained only 5aA or 5bA, etc. Without this information, it was always possible that some other substance in Landis' system (like the steroids he was taking for his hip) was mixed in with the target substance, and was distorting the CIR measurements.

At the first hearing last year, USADA entered "Exhibit 26" as evidence. For some unknown reason, this exhibit was not made public following that hearing. But "Exhibit 26" was made public last month following the CAS hearing. Exhibit 26 is linked on TBV's July 8th "document dump." It LNDD 0333-0345 (or so), and starts at pdf page 340 or so.

It contains over 500 pages of documentation from LNDD, much of it highly insignificant, but as it turns out, the complete mass spec(trometer) data from the GC/MS portion of the IRMS test is there. I've checked with "duckstrap" (some of you will remember him as a real scientist who knows this stuff), and it appears that this complete mass spec data does show that the peaks of interest contain the target substances, and only the target substances.

(This does not settle the "peak identification" issue, by the way.)

Of course, when it comes to anything LNDD does, I am skeptical of the quality of the work. For example, there is no "time and date stamp" on these documents, and they really could have been run at any time on any sample. Like other things, the original ones and zeros must have been erased on that infamous hard drive.

But, for what it's worth, it appears we now have the complete mass spec data for the peaks of interest, and it seems the data supports USADA's case.

syi

jrdbutcher said...

"For example, there is no "time and date stamp" on these documents, and they really could have been run at any time on any sample."

I have not sifted through those many pages and, until I do, I'll take you at your word that the (a)mass spec data is there. With no time/date stamps, it may very well be from a source other than Floyd's samples. If it was from his samples and the prosecuting side could remotely prove so (low standards of evidence), they would have emphasized the point as it would have supported a slam dunk for their case. I'm more than sceptical.

Larry said...

Mike -

Before going further, credit where credit is due. Floyd Landis and his team have been remarkable in the way they have shared information with us. They appear to have given us access to just about everything, whether the information is favorable to Floyd or not. The impression I have is of a man who believes that he has nothing to hide.

I was doing some research today on the wattage figures for this year's Tour compared to previous Tours, and ran smack dab into all of Floyd's wattage figures for 2006. He shared that data with us too, even before the doping issue arose.

Whether you like Floyd or not, whether you think he's innocent or guilty, you have to be impressed with the open and transparent way he's treated this information. When people ask me if I think Landis doped, I like to point to the information Floyd has shared with us, and his insistence that his AAA hearing be open to the public, and I ask whether a guilty person would act this way. But truth is, even INNOCENT people don't tend to act like this. It's simply remarkable, that is all, and I think it says volumes about the quality of this man's character.

All this is introduction to the fact that the information you've pointed out here is certainly bad for Landis. It does not prove that Landis doped. It does not go against any argument made by the Landis team at either arbitration hearing. But it does tend to indicate that there's nothing hidden in those GC/MS peaks that would help explain away Landis' AAF.

I looked briefly at the mass spectra data in Exhibit 26. It appears that this data looks at peak purity by analyzing the ions that were received by the MS at a particular range of retention times. So, for example, if a peak begins at a retention time of 15 seconds and ends at a retention time of 16 seconds, then the mass spectra data for that peak would consist of a count of the different ions received at the MS having a retention time of between 15 and 16 seconds.

So, I have two questions for you, and I'm not sure if there are answers to these questions. First, how did LNDD choose the "slice" of retention times to analyze for each peak? Given the issues we've discussed with TBV and Ali here involving background noise and peak interference, it's not an easy thing to determine when a peak begins and ends.

Second, if there's uncertainty regarding when each of the peaks at issue in the Landis case begins and ends, are we REALLY certain that the mass spectra data in Exhibit 26 proves that these MS peaks are pure?

Finally, you indicated that the LNDD data in Exhibit 26 does not settle the IRMS peak identification issue. I agree, but I think that this data ALSO does not settle the question of IRMS peak purity. Since LNDD did not use the same GC settings for the MS and IRMS portions of its CIR testing, we cannot be certain that IRMS peaks were pure even if the MS peaks were pure. Agreed?

jrd, it is of course possible that LNDD simply made up this evidence. However, there's nothing I see on the face of this evidence that would lead me to question that this evidence is genuine. The mass spectra charts all refer to the Landis sample numbers, and the start and stop times shown for each graph match the general vicinity of the beginning and end points of the relevant peaks on the GC/MS chromatograms. I agree with Mike on the absence of date and time stamps, but if you think that LNDD might have made up this evidence, then they certainly could have made up evidence containing time and date stamps.

("Eightzero") said...

Once again, I join Larry's opinion. It is well articulated.

For me, the issue is not whether Floyd doped. I don't believe he did, if for no other reason that I have not been presented with compelling evidence (any credible evidence for that matter) that he did.

The evaluation of exhibit 26 doesn't change my opinion. While others may be able to say this data is evidence of USADA's charge, I simply don't see that a proper foundation for its evaluation has been made. When I took evidence class, my professor likened the introduction of evidence to that of a "clean room" like you see used to prepare spacecraft before launch. We want to know exactly what it is that has come into that room, so the little particles of extraaneous "stuff" you might find later can be traced. (This seems to be an interesting point for the Phoenix lander team recently. Seems they claim to have found perchlorate in the Martian soil. Odd that this is the same substance used in the rocket propulstion system that got their lab there. But I digress...) I did look to see if the exhibit 26 data was addressed by the defense team. I looked at Arnie Baker's 2.0 Wiki Defense document, and don't find a specific reference. Anyone able to point it out to me?

This case is about the USADA having scientific evidence to support their charge. I don't see that the evidence is objectively sufficient. I beleive extraordinary claims require extraordinary proof.

Five other arbitrators disagree with me. While I'm not surprised others don't share my level of skepticism with scientific matters, I am disappointed that this rejection of science is destroying both sport and Floyd.

Larry said...

For those like me who still turn to TBV for the latest news, the UCI (remember them?) has declared Giro KOM Sella Positive for CERA-EPO.

Mike said...

Larry, I agree with you about Landis' willingness to make everything public. Highly admirable, and in my view (as in yours), a sign of innocence.

Regarding your questions - I'll do my best, which may not be very good:

First, how did LNDD choose the "slice" of retention times to analyze for each peak?

Obviously a good question. The mass spec data we see in LNDD 0333 (and following) covers about nine and a half seconds for the 5bA peak. But that looks about right for the 5bA peak on USADA 0321. It could cover the whole peak (except for the little ramp before the peak and the little bump after the peak). However, USADA 0321 also seems to show that the apex of the peak is at 15.14 minutes. That is not the center of the data set on LNDD 0337, though - it goes from 15.035 - 15.195 min. I have no explanation for that. Intuitively I would think they would center their measurement on the apex of the peak, but maybe that is not how it works.

Ugh, gotta go, more later, hopefully.

Larry said...

8-0 -

Thanks!

I don't think that the mass spectra data was addressed in the Wiki Defense at any point. This was never part of the Landis legal argument. It was much discussed here, and discussed to some extent at DPF.

I think I've said this already, but I no longer believe that the WADA arbitration process, or for that matter any legal process, is capable of functioning as a scientific review board for laboratory work. It's probably worthwhile to have lab work subject to legal review, as this may keep the labs somewhat honest and may catch the most obvious and aggregious lab errors. However, when a lab puts its full prestige behind a particular result, and if the other labs line up in support, what can we realistically expect a few arbitrators to do?

The best we can hope for is to require the labs to be the best labs they can be, and for the rules to be set up in a realistic way that anticipates that labs will make mistakes that the lawyers will be unable to catch. If we have to tolerate some amount of lab error in doping control (and I see no way around this), then the sensible legal response is to lower penalties and reduce or eliminate the stigma of "fault" that accompanies an AAF.

Of course, doping control is moving in exactly the opposite direction.

Mike -

I think that these peaks rise more slowly than they fall, which would explain why their apexes seem to occur close to the end of the period covered by the mass spectra graphs. My guess is that the reported retention time for each peak is supposed to be the apex of the peak, and again you'll see that these reported retention times are towards the end of the periods covered by the mass spectra charts. If I'm right about this, then LNDD could not center their mass spectra measurements around the peak apex.

I've tried to eyeball the start and stop times for peaks on the GC/MS chromatograms against the start and stop times shown on the mass spectra graphs. On this basis, the start and stop times for some of the mass spectra graphs seem a bit off to me. Of course, it's not really possible to determine peak start and stop times with any precision by staring at the chromatograms in the original exhibits.

I DID compare the duration of time covered by the mass spectra measurements for Landis' B sample against the coresponding duration of time covered by the mass spectra measurements for the mix acetate. Note that the time covered by the mass spectra for Landis' 5bA is about 6% shorter than the corresponding time for the mix. This probably means nothing, and I'm not sure it would matter if a little bit of a peak beginning or end got cut off. If the graphs are to be believed, then it looks like they analyzed most if not all of the peak and what they analyzed was relatively pure.

Russ said...

Hi Mike, Larry et al,
Great find Mike!

I am least qualified to analize this stuff but I must say I am amazed at how close the mix acetates are to the sample in each pair as to the peaks position and height. Only the mix acetates all look dirtier. I though the urine samples were supposed to be the dirtier and that doesn't appear to be the case. Also I wonder why the abundance scales match until Fraction 3? Then the sample are on a lower abundance scale for everything.

BTW I wonder how much alike these things normally look? My alarm bell went off for that.

Larry said...

Russ -

Great questions. I'll give you my understanding here, so long as YOU understand that this is an area where Duckstrap is the expert and Mike knows a great deal. If I make any mistakes below, I trust that Mike or someone else will correct me.

The graphs we've focused on most often here are chromatograms, which give a picture of everything that's in a given urine sample. For example, take a look at USADA 0348. This is a GC/MS chromatogram of the F3 fraction of Landis' B sample (the fraction that got Landis into trouble). Every peak you see in this chromatogram represents a different substance in the F3 fraction of the Landis urine sample, with the most important stuff labelled by LNDD. There are at least two dozen substances shown on this chromatogram, though some of these substances are present only in small amounts.

In contrast, what you see at (for example) LNDD 0338 (sorry, I don't have a way to link to this single page, but you can see this page in the larger Exhibit 26 document) is a graph showing the complete mass spectra for a single peak - the 5aA peak - contained in the USADA 0348 GC/MS chromatogram. This is NOT a picture showing everything in the sample -- this is a graph that the scientists can use to identify what's contained in a single peak. The pattern of lines you see in each graph on LNDD 0338 is like a molecular fingerprint -- any substance analyzed by the lab will produce a unique pattern of lines that can be used to identify the substance.

If the lab has done its work correctly, the two graphs you see on LNDD 0338 should be nearly identical. The top graph on LNDD 0338 shows the mass spectra for the 5aA peak that's contained in the lab's test mix, and the bottom graph shows the mass spectra for what LNDD has identified as the 5aA peak in the Landis B sample. These two "molecular fingerprints" are nearly identical -- if they weren't close to identical, then this might indicate that LNDD had not correctly identified the peaks on USADA 0348, or that these peaks contained a combination of substances. In either such case, we'd have reason to question the accuracy of the lab's work.

You raise an interesting point regarding the differences in the abundance scales for the two mass spectra graphs on LNDD 0338. It appears that there was more 5aA in the mix than there was in the Landis sample (more than 3 times as much). If you compare the abundance for the 5aA peaks at USADA 0309 and USADA 0348, you'll see about the same 3 to 1 discrepancy. I'm not sure why this should be -- maybe LNDD was starting to run low on the available Landis B sample by the time it got to the third (and last) fraction of this sample, so it didn't have as much stuff to analyze. But you're quite right -- the abundance scale for the Landis F3 fraction is a lot smaller than the scales for the first two fractions. I wonder if this is significant.

Mike said...

I put the mass spec data in a separate file at archive.org. Here is a link:

Ex 26 mass spec data

syi

Mike said...

Russ, I think Larry said it quite well (but he flatters me - remember, I'm an Idiot about all this).

I also don't know for sure the significance of the difference in abundance. I would think it wouldn't matter, because it is the ratio of the ions, not the absolute numbers, that matters.

And continuing with your earlier questions, Larry:

Second, if there's uncertainty regarding when each of the peaks at issue in the Landis case begins and ends, are we REALLY certain that the mass spectra data in Exhibit 26 proves that these MS peaks are pure?

Based on what I remember, the peak start and stop times wouldn't be so much an issue of for the GCMS part of this, but for the IRMS part. And two things would be at play there. First, based on my reading and what Ali showed with the software, I think the impact of the isotope effect could make some difference to the CIR reading based on the exact start/stop times (the limits of integration). Second, as you said, Larry, even though (if?) they used the same column for both the GC/MS and the IRMS, the temperature ramps were different and thus, hypothetically, stuff could move around (in time, the x-axis) between the GC/MS and the IRMS. And different substances move around different amounts, so something could have been in the peak in the IRMS, even if it wasn't in the peak in the GCMS. And, of course, there is no way to check the purity of the IRMS peak. I think it is VERY unlikely that that "contamination" happened, but I think it is possible.

That last point is frustrating because WMA designed a machine in the late 90s that would make this (and other problems) a non-issue, but LNDD didn't use that type of machine. If I understand right, both this peak purity issue and the peak identification issue would be non-issues if they had used WMA's type of machine.

syi

("Eightzero") said...

Larry, I think your comment "what can we realistically expect a few arbitrators to do?" is spot on. No where in the qualifications for being an arbitrator is the ability to actually understand science. For all we know, the CAS is populated with arbitrators that failed science class.

One fairly simple solution is to prohibit arbitration of these types of disputes. CAS would just love that.

Congress could act to prohibit arbitration clauses in professional sport contracts, and this would at least offer a measure of protection for US athletes. I can see ASO then not ever inviting a US team to their show. But of course, this shines the light even brighter on the main issue: sport at this level is almost wholly politically controlled.

As we will all clearly see starting 8/8/08.

dailbob said...

Eightzero,

"No where in the qualifications for being an arbitrator is the ability to actually understand science. For all we know, the CAS is populated with arbitrators that failed science class"

At one point in time, I suggested to Mike in an e-mail conversation that it might be better if the panels were made up of 3 unconflicted (ideally) Analytical Chemists who routinely run these analyses along with a legal advisor (familiar with the ISL, WADA rules etc), instead of being made up of 3 Lawyers with a scientific advisor. Please note that this is NOT a slam against lawyers (I'd stand in front of Larry, Suh and Judge Hue anyday). My thinking was that in a battle of scientific experts, a panel made of Analytical Chemists couldn't be bamboozled by anybody, and you'd get a judgement based on the merits of the science, which is where the real question is.

Regards

dailbob said...

Larry,
"but I think that this data ALSO does not settle the question of IRMS peak purity. Since LNDD did not use the same GC settings for the MS and IRMS portions of its CIR testing, we cannot be certain that IRMS peaks were pure even if the MS peaks were pure. Agreed?"

Agreed. Although, having said that, the differences in the temperature profiles that they used were unlikely to have caused peak swapping. I've never stated this here before, because it was my own personal speculation, but it looked to me that the difference in temperature ramps was done to try and make the peaks on the IRMS line up with the GC-MS. Much has been made here of the elution time differences between the two machines being due to differences in "plumbing." This made no sense to me, because the "plumbing" is the column itself, which was supposed to be the same on both machines. It seems more likely to me that the difference in elution time was caused by either 1.)the pressure differential at the inlet and outlet of the column being different on the two machines (the GC-MS is under high vacuum and the IRMS combustion must take place at approximately atmospheric pressure) or 2.) the same column was not used on both machines.

The pressure differential could, potentially, be compensated for by increasing the inlet pressure of the IRMS column to try and get the same pressure differential (not easy). If you did this, it might make sense to try and tweak the temperature ramp to get the peaks to elute at the same time(s).

Again, the above it personal speculation, so take it with a grain of salt. This just made more sense to me than the plumbing statement.

regards

Larry said...

8-0, d-bob, maybe I need to be more clear on this point. I don't think that there's much that three arbitrators can do to review a lab's work on a given case, even assuming that the arbitrators are reasonably expert in the science. There's just too much that goes on behind the closed doors of the lab, outside of the public view. The lab's documents seem (to me, at least) to give us barely a glimpse at what the lab really did, and what the lab really found. I don't see how we can reasonably count on a lab to faithfully and accurately document all the ways it screwed up a particular test. If the lab fails to perform the test correctly, why should we expect that the lab will faithfully report its failure to us?

There's no substitute for good labs doing good work, and there's no legal process that can protect us against bad labs doing bad work. If I was put in charge of fixing this flawed system, I'd focus 99% of my effort on improving the labs.

d-bob, I understand from the testimony before the CAS that even the Landis experts did not believe that the differences in the temperature ramp would cause peak swapping. All I'm stating is that the differences in the temperature ramp could cause peaks to move, or could account for greater interference in one test than in the other, or even for peaks that were separate in one test to fail to separate in the other test. Yes, I understand from the experts that none of this is "likely" and that it's "likely" that pure peaks in the GC/MS translate into pure peaks in the GC/IRMS notwithstanding the difference in the temperature ramp. But I grow weary of looking at the evidence in this way. I'd rather rely on the lab's adherence to a sound procedure than having to be confronted all the time with the "likely" meaning of the lab's (IMHO) flawed methodology.

d-bob, I've performed the following thought experiment. I order a Diet Coke at McDonalds, and the guy behind the counter gives me a straw that is 120 feet long and 1/100 of an inch in diameter. My guess is that no matter how hard I suck on that straw, I'm not going to get any Diet Coke. My guess is that no matter how hard I blow into that straw, I'm not going to be able to blow any bubbles into that Diet Coke. Yes, I've read about the pressure difference at the end of the GC-MS column versus the GC-IRMS column, but with a "straw" this long and this narrow, why would the pressure at the end of the straw amount to anything?

Your explanation is the best I've heard for why the LNDD used different temperature columns, but it doesn't add up for me.

dailbob said...

Larry,
"All I'm stating is that the differences in the temperature ramp could cause peaks to move, or could account for greater interference in one test than in the other, or even for peaks that were separate in one test to fail to separate in the other test"

I still agree with this statement. The point I was trying to make was that what they did with the ramp looked like more of a purposeful "tweak" to me than a mistake. Having picked up a little knowledge of GC "by osmosis" over the years, I'm asking myself "what the heck would you do that for?" (especially when I think that maintaining the identity of a compound as you move from the GC-MS to the IRMS is a big enough challenge as it is). What I wrote was the best explanation that I could come up with. Again, just my musings.

My masters degree is in engineering, and I want to tell you your thought experiment is a great fluid mechanics example! While it seems that the column is long with a tiny orifice, they are engineered to make sure you get some "diet Coke" out of the end. Otherwise, there would be nothing eluting from the column. Also, the Agilent software allows you to determine the pressure at the outlet. I'm going to visit my Analytical Chem buddies today to determine what flexibility it has in allowing you to vary the inlet pressure (beyond what's pre-programmed). The pressure differential is important, because it will definitely effect residence time in the column.

"I'd rather rely on the lab's adherence to a sound procedure than having to be confronted all the time with the "likely" meaning of the lab's (IMHO) flawed methodology."

One of the things that really bothered me is that it didn't seem that Landis could get the lab SOPs and validation report. The FDA expects me to have SOPs where I state what I'm going to do, and to document that I did it. If I didn't document it, it didn't happen, in their eyes. Your desire to know if the lab adhered to a sound procedure could be satisfied in this way. But, not being able to get their SOPs completely frustrates this. From my perspective as a "lab guy", not being able to get their SOPs is tantamount to not being able to see all of the evidence against me (I'd like your, and the other Lawyers here, thoughts on this).

Regards

Russ said...

Larry and Mike, Thanks esp for the clear explanations! I am at the age that if I don't review this stuff frequently, it begins to slip away and I forgot exactly how the mass spec fit into the gc/ms picture.

And d-bob, regarding the temp ramp and pressure, sorry I can't recall the source but the pressure was, by design, supposed to be in spec, I am rather sure, which would make an intentional variation to be a non standard use of the machine.

As to the temp ramp, as I recall, the irms ramp was a good deal faster than the gc/ms ramp. The tech sources warned against this in order to achieve proper seperation. So if it was intentional, it must have been either to save time or muddy the evidence.

Regards,
Russ

dailbob said...

Larry,
In the last paragraph of my last post, I mention not being able to get validation reports and then figure out that I missed LNDD 0456 (because I'm not patient enough to wait for the download). So, today, I wrest the "idiot" title from Mike (which I should take permanently)! Apologies!