Friday, November 30, 2007

Different Columns!? Hiding in plain sight

In comments, Mr. Idiot has noticed something that was in plain sight all along, if anyone had noticed.

The Majority wrote, in the key seven paragraphs, at paragraph 188:

The GC column is, of course, the same in both instruments.

This is a "misstatement of fact" by the Majority, on which some significant conclusions are based.

We see that the columns were NOT the same, based on parts of USADA 124, 153 for the A sample, and USADA 303 and 329 for the B sample:


USADA 124: Using an Agilent 19091s-433 for the GC/MS.

USADA 153: Using a DB-17 (as did Shackleton) for the GC/C/IRMS

USADA 303: Using an Agilent 19091s-433 for the GC/MS.

USADA 329: Using a DB-17 for the GC/C/IRMS

These are completely different columns, with different "polarities". Polarities are known to causes significant location changes in peaks, including changing the order of peaks. It isn't known as fact whether these columns change the order of the 5bA and 5aA peaks in the setup used by LNDD.

For the record, Agilent used to be part of HP, and now also owns the rights to the DB-17. So both are made by the same company, but are of completely different specification.

Mr. Idiot writes:
[T]his all comes from Arnie Baker through email. I do have explicit permission to make it public. The down and dirty version of it is this - most of it is right there in the original LDP:

USADA 0124 says that the column used in the GCMS was an "Agilent 19091s-433." If you dig in the Agilent website, it is clear that an Agilent 19091s-433 is an "HP-5ms" column.

USADA 0153 says that the column used in the GC/C/IRMS was a "DB-17ms." Again at the Agilent website you can find the DB-17ms and learn that it is a different column with different characteristics than the HP-5ms.

Just as one example, the HP-5ms is "non-polar," and the DB-17ms is "mid-polar."

Baker has shown that the LNDD SOP required that the DB-17ms be used for both GC/MS and GC/C/IRMS, but for some (undocumented) reason they used the HP-5ms for the GC/MS. The evidence of that is a little more complicated, so I'll wait on explaining it.

The information Baker has given me also shows specific examples of how substances elute at different times and in different orders on these two columns, and even has an example of a steroid doing so (although not specifically 5aA or 5bA). That's part of the stuff that is graphics heavy and TbV will have to do something with it.

Obviously one of the first things you learn about the connection between GC/MS and GC/C/IRMS is that the chromatographic conditions have to be the same. Now we realize that not only were the temperature conditions different, the columns were different too.


UPDATE: more in Different Columns - Details.


Lowryder said...

Wow. Information that got through despite the "disappearing" raw data and whiteout. As you said, it was hiding in plain sight, from the LNDD "data sanitizers" as well as everyone else.

Larry said...

Before saying anything else ... I think this is the single most damning fact that has emerged to date against the case made against FL by USADA and LNDD. I have not done a complete analysis ... but at this point there may be nothing left of the case against FL.

It is very strange, LNDD did the exact same column switch for both the "A" and "B" samples, in tests many days apart. This looks like their unofficial standard operating procedure (from what Mr. Idiot posted, it looks like their actual SOP required the use of the same column in both the GC/MS and GC/IRMS tests).

Does anybody have any idea why they might have done this?

Do we have this same information for the testing of the other FL "B" samples? Did LNDD switch columns for these tests as well?

("Eightzero") said...

How I hope the CAS shoves this right up LNDD and ASO's you-know-what.

Imagine CAS ruling for Floyd. Now picture ASO making a call to Oscar Peiro. said...

It's not a 'switch', it is two different instrumets with different GC units having different columns.

It appears to be the house setup also used in the other B testing.

Why? Dunno. Maybe they thought they had better separation this way. Or maybe one day they replaced the column in the MS without realizing they got the wrong one. Or maybe they use the ms for another test that needs the agilent and they forgot they need to change it.

I don't have the accreditation dogs; will try to get them.


Larry said...


Did the accreditation dogs provide 128 bite protection for LNDD's network?


m said...


The B sample GC's are shown at USADA 0329 and USADA 0303.

The size of the columns are the same: length 30 meters, diameter 25 mm.

I wonder if it makes any difference whether the models are different? We'll have to wait on more definite information.

From your comments it sounds like LNND used this same set up for both A and B samples, and possible all the retests. If so, this suggests it was SOP.

Does the fact that the Landis camp didn't raise this issue suggest anything to you? -)

Larry said...

M, I think the FL team flat-out missed this fact. USADA's argument includes a statement that the GC was the same for the MS and the IRMS; Suh would not have missed an opportunity to contradict USADA if he'd known about this fact.

In preparing for trial, I think that FL's team thought that RRTs would be enough to invalidate the GC/MS - GC/IRMS comparison. If they knew that Brenna would testify that RRTs don't matter for this comparison, then they might have dug deeper to find facts to contradict Brenna.

I also wonder what Brenna would say if confronted with this fact.

I'm going to spend some time this weekend looking at this. But the use of different columns (or as TBV says, the use of different GCs) to run the MS and the IRMS appears to me to devastate USADA's position on a number of points, including specificity and peak identification. The use of different columns allows for the possibility that peaks DID change position from the MS to the IRMS. We don't know what peaks might or might not change position, but if any peak changed position, it could screw up the pattern matching advocated by Brenna.

The GC switch also means that whatever LNDD might have done to check specificity on the MS would not prove anything for the IRMS.

There are issues involving LNDD's departure from its written SOP.

I'm trying to be careful here. Maybe there's something I have not considered. I'm keeping an eye on DPF to see if this topic comes up over there.

cclarke said...

Why do you think the GC instruments would procude different results?

m said...


Right now I don't think different results were produced even if different GC instruments were used. Although Landis partisans are trumpeting this as a possibility.

There is no evidence the 5A and 5B peaks switched.

The GC-IRMS 5B peak in the Landis F3 sample is anchored because it has the same retention time as the IRMS Cal Mix 5B and the IRMS Mixed Urine 5B.

Moreover, the pattern (visual and retention times) of the 4 central peaks that include the 5A and 5B match in both the GCMS and GC-IRMS, despite any difference in model of the GC columns.

So it is almost impossible that the 5A peak switched with one of the other 4 central peaks, not including the 5B which we know is anchored. Visually the sequence of peaks stretched out in the GC-IRMS when compared to the GCMS. They did not switch.

BTW Larry you should check and see whether the majority said the GC machines were the same, or just the GC columns. The columns are the same size, although I don't know about the other specifications. said...

For sake of argument, I might spot you that the three main targets won't move.

However, I don't know what that does to LNDD's position with regard to the validation and accreditation of their methodology. How can they claim to meet a specificity requirement if the method was validated using a DB17 for the GCMS but they used the Agilent with different polarity?

I'm not sure I see how it can be dismissed legally.

TBV said...

That is -- other peaks might move, into co-eluting locations, and how would you have any idea if that happened?


BustinBilly said...

You won't get peak swapping between the DB 5 and DB 17 columns.

neb said...

On what basis do you make the claim that you don't get any peak swapping between DB5 and DB17 columns. My (albiet limited) experience with GC is that any time you change the polarity of the column, you get variations in of the order of peaks. With a non-polar column molecules basically seperate by their atomic mass. However, with a mid-polar column if the molecule has any polar regions they stick a little bit to the polar column and they come out of the column later. With a biological sample such as urine, there are plenty of polar regions expressed.

Mike Solberg said...

I'll try to explain the departure from the SOP, based on what Arnie Baker gave me.

The names, or titles, of the relevant SOPs are in the original LDP. In English the name for an SOP in the documents is "Method Information," as on USADA 0124, - specifically on that document it is "M-AN-52." In French they are called “Codification,” as on USADA 0153 - specifically on that document it is "M-AN-41."

For the IRMS, USADA 0153 is a standard "mode of operation" document for this part of the process, and it shows that for "codification" M-AN-41 the proper "colonne / column" is the DB-17ms. But there is no comparable “mode of operation” document for the same part of the GCMS process. We know it was SOP M-AN-52, and we know that what was actually used was the HP-5ms.

Of course, LNDD did not, and was not forced in discovery to, reveal the actual SOP documents. So why does Arnie claim that the use of the HP-5ms violates the SOP?

Because in the reprocessing documents we find a bit of the content of M-AN-52. On LNDD 0664 we see a “mode of operation” document that shows that the DB-17ms is the proper column for M-AN-52 (the same for the GCMS). What they actually used was the HP-5ms. So it was an undocumented violation of their own SOP, and thus of ISO 17025.

TbV, maybe you could link to the right document?


Russ said...

"My (albiet limited) experience with GC is that any time you change the polarity of the column, you get variations in of the order of peaks."

Would you care to embellish your excellent explanation with the effects that the different temp ramp rate and pressures would have on the elution,with these columns as well?


Larry said...

cclarke, I'm not sure what you mean by "different results". Are you asking whether FL's test results would have been different if the testing lab had used a different GC machine? The answer, I think, is "maybe", or "it depends on which machine". If you're asking whether the results in the FL case would have been different if the testing lab had used the same GC machine for both the GC/MS and the GC/IRMS portions of the CIR test that ultimately led to Landis' conviction, the answer is "we're not sure yet", or perhaps "we'll never know."

But if you're asking whether different GC machines behave differently, whether they can present peaks in a different order, or a different pattern, the answer is "yes". If you're asking whether different GC machines have different abilities to separate substances found in a complex mixture such as an athlete's urine sample, the answer is "yes". If you're asking whether the selection of a GC column is a critical component of the design of a drug testing system, the answer again is "yes".

If you're asking whether it's important for a lab to maintain consistent chromatographic conditions when it performs the GC/MS and GC/IRMS portions of a CIR test, the answer is "yes". If you're asking whether the selection of the GC column is the single most important of these chromatographic conditions, the answer is "yes". If you're asking whether any competent lab would intentionally use different columns for the GC/MS and GC/IRMS portions of a CIR test, I'd say that I'm not an expert and I'm willing to be corrected, but that I think the answer is "no". said...

I advise moving comments to Different Columns - detailed, which offers an example of significant ordering changes between a DB-5 and a DB-17.