After reading TBV’s report of the Floyd Fairness Fund event in San Francisco, I was very intrigued by Dr. Arnie Baker’s comments about the mass spectrometer being run at too high a pressure. But since TBV didn’t have a photo of the slide which detailed Dr. Baker’s comments, I wasn’t sure what exactly Dr. Baker was referring to.
Following a hunch, which later proved to be wrong, I asked one of my sources what the implications of running a mass spectrometer at too high an internal pressure. You can see the response to that question here. TBV’s comment on the post I subsequently wrote said:
I understand the improper ion separation argument, and look forward to deeper analysis. At the same time, I wonder if it might also be completely innocuous, and arguably an improvement in the technique recommended by the manufacturer. You’d want documentation to show that it was an improvement, really, but it might be. It seems to me possible for the pressure to be like a “gain” knob on an amplifier. Adjusting the pressure might bring things into a more accurate or convenient range on the column. If it separates correctly, and is calibrated against known samples accurately, then it might work better than the original spec. We don’t know without the calibration data deBoer requested and has never received, and possible examination of the electronic data files rather than printouts.
On the other hand, it might be like setting the amplifier to 11.
Originally, the post Under Pressure was a response to TBV’s question, going into the physics of how running the mass spectrometer at too high an internal pressure might cause incorrect results. At the FFF event in San Jose a couple of days later, TBV got an image of the slide where Dr. Baker details the pressure problems.
Seeing Dr. Baker’s slide and what it had to say changed my understanding of the problem being alleged. (I’ve saved the text of the original Under Pressure for use later, should the need arise.)
The Penning gauge, being used to measure vacuum pressure, was recording pressure at a certain point in the system, and the manufacturer’s documentation that Baker cites shows that at that point, the pressure should have been lower. The manufacturer’s documentation says that a pressure above a certain amount indicates “major leaks.”
The operating pressure the gauge recorded, as shown in an LNDD document cited by Dr. Baker, is higher than that threshold. According to the manufacturer’s documentation, the system should have been shut down and the leaks fixed before continuing. You can read my conjecture about what the implications of leaks in the system might mean in the Under Pressure post.
That conjecture caused a bit of a ruckus over at The Daily Peloton Forums, where Tom Fine put a link to the article. Two of the people who post regularly over there (D-Queued and rationalhead) were not pleased with the article, and rationalhead went so far as to call it a “myth.” But in a later comment, someone who goes by the online moniker “IandC,” responding to one of Tom Fine’s comments, had this to say:
Tom, I am very familiar with IRMS.. I did not read all the links and you referenced and I can only comment on some general issues. You are perfectly right that contaminants from maintaining poor vacuum could be a serious concern. I could even say that proper vacuum is one of the three most common problems as regards GC-C-IRMS method [italics added] .. They are: I) poor system linearity. II) incomplete combustion III) poor vacuum/leakage. Even a very small leak can lead to seeing duplicate peaks and unsettled analytical runs due to unstable flows. [italics added] This, in turn, can lead to incorrect quantitative analysis due to distorted peak shapes and area.
If I’m following IandC’s comment correctly, duplicate peaks might be interpreted to mean that something was present when it might not have been been there. Or perhaps more of something was present, depending on the nature of the fragments being measured and what those fragments contained.
In mass spectrometry part of the measurement has to do with when an ion reaches the detector. Heavier ions should get there later than lighter ions, all things being equal. Carbon12 and carbon13 differ in mass by the weight of one neutron, which is not a great deal compared to the difference between the mass of a carbon atom and an oxygen atom (or a nitrogen atom or a hydrogen atom), for example. So, in theory, a carbon13 atom would arrive at the detector shortly after a carbon12 atom.
So if you can completely break a molecule apart and have only its component elements being analyzed, you can determine how much of each isotope of carbon is present, based on the peak shapes and sizes.
But if you have a duplicate peak in the region where carbon should show up, that could throw off your interpretation of what’s present. You might find more carbon13 than expected, for example. Which is exactly the kind of thing that IandC describes.
IandC went on to describe a number of measures that should be in the system to keep leaks in check and concluded by saying:
I can only say it would be inconceivable that a lab would proceed with wrong vacuum given the number of checks I described.
It does seem hard to imagine that an accredited lab would do such a thing. But Dr. Baker’s revelation that LNDD didn’t have the operating manuals and documentation for their instrument is equally hard to imagine. So I can’t put it out of the realm of possibility that they were operating the mass spectrometer in a manner not in accordance with the manufacturer’s recommendations. In fact, that seems all too likely.
One of the questions that is not yet answered is whether the documentation shows this operating pressure for one individual run of a test during analysis of Landis’ sample, or whether this is how LNDD routinely operated the equipment. Also missing from the discussion is LNDD’s calibration documentation for the mass spectrometer.
Dr. Douwe deBoer, who was Landis’ observer for the B sample testing requested this kind of documentation, but it was not included in the lab documentation pack forwarded to Landis by USADA. It would be in LNDD’s interest to provide such documentation, if it exists, to bolster their claims regarding the accuracy of the tests. Of course, if there were problems with how the tests were conducted, LNDD would have equal incentive to “lose” those documents.
A mitigating factor in all of this for LNDD is what Wikipedia has to say about the calibration and accuracy of Penning gauges, which is:
Thermal and ionization gauges read pressure indirectly by measuring a gas property that changes in a predictable manner with gas density. Indirect measurements are susceptible to more errors than direct measurements.
The Penning gauge is an ionization gauge, which means it determines pressures indirectly, hence it is more susceptible to error. That could work in LNDD’s favor, if they can show that the gauge in question was consistently wrong and that the nominal reading recorded corresponds to a lower real value.
Sort of like if you calibrated your cycle computer incorrectly. The speed and distance it records would be incorrect. But the error would be consistent, so if you knew the magnitude of the error you could correct for it.
What I got out of the slide from Dr. Baker’s presentation regarding pressures was that the pressure indicated major leaks in the vacuum according the manufacturer’s documentation. This is much different, in my estimation, than what TBV asked about in regards to turning up a gain switch, or turning the volume up to 11. IandC’s comment suggests that the leak could cause problems with the acquisition and interpretation of the data, if the proper measures weren’t in place to keep those leaks in check.
Whether LNDD has such measures in place may become known as more documentation and information becomes available. If, however, the issue Dr. Baker raises applies to more (or all) of the GC/IRMS tests run on Floyd Landis’ samples from Stage 17, then the case against Landis will almost certainly be in jeopardy. Time will tell.
Note: Props to swimyouidiot for reminding me about IandC’s comments in The Daily Peloton Forum. Most appreciated.
After reading your analysis, a thought popped into my head ….
In Landaluze, the same technician performed the a- and b-sample analysis so the case was thrown out. The LA Times says the same thing might have happened here, except this time it was the same two technicians on both a- and b-.
Here’s my thought, the LNDD only has one, or maybe two, that know how to use the IRMS machine. They have been using it wrong for a while and no one else in the lab is in a position to know this.
If indeed the number of qualified people is one (maybe the second in the Landis case was being trained), then one can see how they the mistakes described here could very have been made. All they know is what the previous technician taught them, which was incomplete. And like tales from your elders, information/knowledge gets lost everytime the story is told to the next generation. So along the line, the manuels were lost, someone forgot to tell them to update the software, and/or the information about the proper pressure settings was not passed along.
Do we know the name of the lab technician in the Landaluze case? Is it one of the two lab technicians named in the LA Times?
Do we know how many IRMS tests the LNDD does in a year? If so, is it low enough to say that one or two full-time lab technican(s) could have done all these tests? If it is one, who trained them? Was it another lab technician that was leaving the lab so they sat down with the current technician, showed them what to do, and quit?
(Side note: I see to remember an issue of the b-sample having to be done before the lab closed for its August holiday. I remember thinking this case was so important that they should have made an exception if it got to this. Maybe the problem was this one/two technician(s) was going on holiday, was over-worked like the arbs in the Landaluze case noted, and was having none of canceling their vacation (yet again?) to look at some American’s pee. Especially if the Lab was not going to reimburse them for their holiday plans if they asked them to cancel.)
Again, thinking out loud. Thoughts/ideas?
ORG,
I was thinking about some of this before reading your comment. There was a rush to get Landis’ B sample done before LNDD shut down for their August holiday. Probably not a good idea to rush these tests, because when you’re in a hurry it’s easy to make mistakes because of the pressure you might be feeling, if for no other reason. I haven’t seen the name of the technician in the Landaluze case who worked on both the A and B samples. Perhaps that info is out there, somewhere. If so, I’d like to see it. I don’t know if the lab has a written procedure for running the IRMS process, but my understand of ISO standards is that they should. And they should be able to produce it on request. As to whether their personnel would need special training to run the equipment, and whether only a few in the lab were qualified, I don’t know. But such things should be documented by the lab if that’s the case. I can very easily imagine a situation where only one or two persons qualified to run the equipment are present, and there’s a lot of pressure to do the tests, and someone decides to go on with the test anyway, despite the fact that doing so would violate some standards. I’ll elaborate on that scenario in a post in just a bit …
– Rant
ORG & Rant,
Keep in mind that the testing of Landis A & B samples were prior to the Landaluze case ruling. Therefor I’m going with LNDD Standard Operating Procedure to have these 2 technicians to do the grunt work on both A & B samples of any test. Unless the second was brought in to finish it up before holiday.
Could they have been trying to make a change for the better by training a second tech, knowing what was happening in the Landaluze case? Highly unlikely since most of the other hearings are behind closed doors, including Landaluze, and Landis is the only one to show us information regarding the case. I doubt LNDD saw it coming.
I think chances are very great that most, if not all, B samples of any positive or negative doping case at LNDD can be thrown out. Due to Tech’s confirming there findings. They’ve been sloppy with everything else, why should we assume any different other than mind exercises to run different scenarios?
I would also guess they have there ISO Procedures manual in place, but I’d bet you would have to take some Allegra D to help with the dust allergies before pulling it off the shelf.