Wednesday, September 26, 2007

Discussion about Relative Retention Times

Over in the thread about visual identification of peaks (Discussion with BustinBilly), there have been some comments that lead naturally into a discussion about the applicability of relative retention time across two instruments.

As we know, Paragraph 185 of the Majority Award declared that it wasn't possible, therefore Landis' complaints about the RTTs not being within spec of TD2003IDCR didn't apply.

In the discussion of visual identification, I tried to show that some of the math used in paragraph 185 was incorrect:

M wrote:

Indeed, the majority decision in their mathematical example seems to suggest that what TBV is doing can never result in a match.

They say the retention times of the IRMS GC take longer than the GC-MS machine by some constant amount of time. They use the example of a relative retention time ratio of 10 to 5, becoming 11 to 6, after you add a 1 minute constant.

Well if in fact the IRMS retention times are off by some ADDITIVE constant relative to the GCMS machine, then you can't correct for that by MULTIPLYING in some proportionality correction. By stretching the scale I assume TBV is doing some multiplying. At least that is how I understand the example and the math.

TBV wrote:
The panel got their "math" example wrong. If they'd figured it out correctly, the example they gave as "broken" worked just fine. You are supposed to subtract the fixed offset before you do the multiplcation, so (11-1) / (6-1) works out to 10/5 just like they said it didn't.

M responded:

What is the authority for these two statements:

1. "You are supposed to subtract the fixed offset before you do the multiplcation"

2 "and then scaling (doing the proportional multiplication)." (I asked before, on what basis are you claiming you are entitled to do proportional scaling?

Kevin answered:

You can google "relative retention time chromatography", and the first link you find (http://www.iupac.org/publications/analytical_compendium/Cha09sec237.pdf) will take you to a pdf document entitled "9.2.3.7 Retention Parameters in Column Chromatography". It has a section (p.2) describing the Adjusted Retention Volume (Time). It then goes on to discuss Relative Retention (r) on p. 4. For those who are interested, IUPAC is International Union of Pure and Applied Chemistry. This is pretty standard stuff.
Another reference I've found is a tutorial at the Shimadzu web site, unfortunately in Flash.

(also some interesting stuff at scientific.org)
Although many consider GC/MS to be the "gold standard" in scientific analysis, GC/MS does have some limitations. Because great faith is maintained in GC/MS analysis, erroneous results are not expected and hard to dispute. However, false positives and false negatives are possible.

I'll confess that I've never quite understood some things about RTT. It's been represented to me that it does solve the problem in question (identifying peaks) across different runs, which might include different machines. The Majority of the Panel asserts this is not the case.

Let's explore together the same way we did with "visual gestalt". I'll post some things as I understand them, then comments can react and tell me what I've gotten wrong. I'll collect the ones I think are most relevant here and we'll see where we end up.

Before going on, commenter "M" writes, quoting the iupac reference above:
On the other hand, the unadjusted relative retention (and "relative retention time") values are only reproducible within a single chromatographic system."

There's an implication that RRTs are therefore not valid, except that we are dealing with a single chromatographic system, despite various caveats:
The GC column is, of course, the same in both instruments

-Majority award paragraph 188

So, we'll work through it as best we can and see who is hot-dogging who.

Let's illustrate some terms, shamelessly borrowed from the Shimadzu tutorial, also from an Agilent document on terms.



Figure 1: Basic Parameters

Time zero is the injection, then there is plumbing delay (see majority award para 184).

The "Retention Time" is from injection to the peak detection.

The unretained compound peak seems to be time at which a test substance that won't be delayed by the column is injected until when it gets out. It goes through the column as fast as anything possible can, and becomes a fixed reference for this run.

The "Adjusted Retention Time" is the retention time less the hold-up time.


Figure 2: Relative Retention

In the simplest case, we have two sample peaks, a known reference, and a target that follows. The RRT is the ratio of the adjusted retention times of the target and reference, and shown as the "alpha" symbol.

The figure also introduces K symbols, which are Retention Factors. The retention factor for a peak is the adjusted retention time divided by the hold up time. As a special case, the retention factor of the unretained peak is defined to be zero, to avoid a divide-by-zero.

Figure 3: Compound identification by Time

Figure 3 shows how retention time is used to identify compounds. Given the times observed in a sample with known substances, the times can be compared with times of unknowns in a target sample.

This is how we know in the GCMS that the labeled peaks are what they are -- there was reference run in the GCMS with standards that identified the substances in question, and the retention times there were matched within TD2003IDCR of the Landis target samples.


Figure 4: Example matches of USADA 309, Cal Mix Acetate and USADA 321 Landis F3 f.

Figure 4 shows the kind of match done in Figure 3 with the LNDD calibration sample and the Landis F3. The target peaks line up well, within TD2003IDCR. We note at the left edge by the known standard that the circled peaks of the standard in the cal mix acetate is very different height than it is in the Landis sample. If we'd gone on peak heights alone rather than look at the times, we might have lined the Landis next peak to the right with the cal mix.


Figure 5: IRMS chromatograms from '474 B: Cal Mix, Blank F3, Landis F3.



[ coming, cal mix in the IRMS and the Landis F3 ]



10 comments:

Bill Neumann said...

After spending way too much time on this topic, I've come to the conclusion that what the majority and other supporters are essentially saying is the following:

1. If the GC/MS identifies the substances of interest in a particular pattern of peaks, and
2. The GC/C/IRMS identifies at least one peak that's shows an exogenous Carbon source (C13) in excess of the allowable maximum, and
3. That peak and others in the GC/C/IRMS plot correspond ROUGHLY in relative position and relative height to the peaks of interest in the GC/MS plot, THEN
4. We will assume that the exogenous Carbon came from one of the substances of interest-and the math (RTT calculations that are subject to inter-device measurement differences)is immaterial because we don't have any other reasonable explanation for the C13 exogenous carbon source.

Cub said...
This comment has been removed by the author.
m said...

"Unadjusted Relative Retention (rG or αG )

Relative retention calculated by using the total retention volumes (times) instead of the
adjusted or net retention volumes (times):

[missing equation]


Relative retention (r) and separation factor (α) values must always be measured under
isothermal conditions. On the other hand, the unadjusted relative retention (rG or αG)
values may also be obtained in programmed-temperature or gradient-elution conditions.
Under such conditions, the symbol RRT (for Relative Retention Time) has also been used to describe the unadjusted relative retention values.

Using the same stationary and mobile phases and temperature, the relative retention and separation factor values are reproducible between chromatographic systems.

On the other hand, the unadjusted relative retention (and "relative retention time") values are only reproducible within a single chromatographic system."

http://www.iupac.org/publications/analytical_compendium/Cha09sec237.pdf

Requires further elaboration and investigation

bostonlondontokyo said...

TBV, I know you're mercilessly editing comments about the analysis of the results, but I have a related comment (request) - Is there anyone reading the blog who could help to place the reading of these results into some context for comparison? Is there another kind of test/result (physiological, biological, chemical) that is used in a more every-day context that would help us to understand how these results are interpreted? For example - how would analysis of a OTC pregnancy test compare to analysis of these results?

I'm just finding it hard to know if we're dealing with really obscure details, or major, obvious inconsistencies.

Thanks much.

tbv@trustbut.com said...

BLT,

The OTC test shows pink or blue.

Imagine if the result depended on the exact shade of pink, as seen under a certain kind of light as viewed from a certain angle. The debate here is kind of like deciding if the light is the right one, the angle is correct, and whether the one we're comparing it to is being observed in the same conditions or with a different light and angle.

Another difference is that in an OTC test, if we'd waited this long to know the answer, we'd be feeding and changing the result, because there is a clear confirmation of the indicaton, and validation by millions if not billions of tests in the field.

Here we have only the interpretation, and non-standard tests -- each lab does it differently. LNDD has run on the order of 500 of these tests in total, and declared around 30 of them positives. It isn't a large sample, and the study that ran that showed they did detect doping didn't show whether they also detected known non-dopers.

TBV

bostonlondontokyo said...

TBV, excellent and clear analogy - many thanks...

To add another question to the 'pink or blue' results interpretation, tell me if this is accurate: Would you say that the USASA has a standard that simply says 'If it's pink, it's pink. That's all we need to know.' If this is the case, I can see why Landis' case result has become so conflicted, as seen in the arbitors' different decision. One can easily argue that Pink is Pink - just as easily one can argue, 'Where is the standard that defines what Pinkness is?' - would this be an accurate assessment? Sorry if it seems that I'm dumbing down the discourse here, but it's helping me to understand the basic situation.

m said...

TBV

"There's an implication that RRTs are therefore not valid, except that we are dealing with a single chromatographic system, despite various caveats:"

1. Not clear that we are dealing with one chromatographic system, even though GC columns are the same.

2. "RRT" here is a term of art and is equivalent to "unadjusted relative retention". It is not all clear to me that "unadjusted relative retention" is what was used in this case. Rather the terms "relative retention" which is defined in another section may be what was used, because there was a need to adjust for the time delay between the GCMS and the GC-IRMS.

3. For the results to be reproducible, the temperatures, mobile and stationary phases must be the same. We've been told that the temperatures were not the same.

tbv@trustbut.com said...

M, I'm moving there slowly. It's suggested that the things you cite can be corrected for using standard methods, but it'll take a bit to see how sensible that is.

TBV

Mike Solberg said...

tbv, I think your "(b) that we can map the GCMS to the IRMS in some way" is what bustinbilly here, and OMJ on DPF, are both saying, isn't it? I'm not sure I followed your earlier discussion with the charts of the peaks (I couldn't always see which peaks you were talking about), but aren't people claiming that the pattern of the peaks in the IRMS lines up just fine with the pattern of the peaks in the GCMS? Do you think you have come up with a satisfactory rebuttal to that?

BTW, I'm not much of a conspiracy theorist (I tend to suspect incompetence, or even just lack of precision, rather than malfeasance), but isn't it interesting that they had TWO ways they could have confirmed the substances in the IRMS (the complete mass spec data, or the available calibration sample) and we don't have either one? And so we have to rely on "eyeballing" it?

syi

Mike Solberg said...

tbv edited out the (b) I mentioned above, so that might not make much sense now.

tbv, it sounds like you might successfully get to the end of this and still be faced with OMJ's question. What else could possibly be causing the extremely negative 5a number? If you want to claim it is not from the most obvious thing (exo-T), then isn't it reasonable to expect you to come up with some other suggestion of what it could be - something with a very negative CIR?

I'm rootin' for ya. But the rabbit hole is deep.