The world's greatest sporting event is on at the moment; I'm sure you are all glued to the live feed from cyclingnews.com every morning (in the US, whatever time wherever you may be) [Go Vande Velde!!!!!]. You may possibly have missed the expected, i.e., the first doping positive- Beltrán was booted for suspicion of EPO doping. And you know we just love the doping stories around here. [Update 7/16/08: Another EPO
positive non-negative "A" sample]
One of the larger ones of recent interest was the Landis debacle. The debacle in which the accused rider's defense rested in no small part on the probity of the testing lab and the validity of the analyses which provided evidence of his testosterone doping.
This brings me to a fantastic and highly relevant post that is up on the 49 percent blog. In a recent entry, Samia details some of the difficulties of producing ready-for-the-courtroom analytical results. [Update: On re-reading, I didn't really make it clear that Samia is not writing specifically about sports-doping but rather about analysis of soil and water samples. The essential principles having to do with experimental / analytical design and interpretation of results are generalizable to many other types of chemical analysis]
Because this kind of analysis involves rather small minimum reporting limits and use of very sensitive instruments (GC/ECD), there is a lot of quality control involved. This data has to be defensible in court if required, and auditors are very picky about how things are done (with good reason). You can't just stick the samples on the GC and report the raw data without comparing it to controls. Examples of possible variables are things like matrix interference, column contamination, inconsistences in sampling methods, possible contamination of reagents or glassware, etc. So there are all kinds of what we call "blanks" to control for this interference and help nail down problems if we start getting weird numbers somewhere. We spike these samples with known amounts of the analyte of interest, as well as another compound called a surrogate that can (again) help weed out where problems are occurring. It works like this: if you know how much spike solution or surrogate you injected into a blank, and you get some weird percent recovery, it's time to investigate possible lab contamination. There are also bottle blanks where the engineers in the field rinse their glassware and send the liquid rinse as a separate sample-- this is a control measure that evaluates possible field contamination.
She is much more capable than I of describing the complications, of course. My experience with analytical chemistry is mostly through collaborations. Nevertheless I learned very quickly that my view of "Gee, can't I just hand you a bunch of samples and you can give me nice reliable, repeatable levels of exogenous drugs, neurotransmitters, their respective metabolites? Three different tissues from two different species? All by next Thursday? ..thx k bai!" was......naive.
Even before you hand off experimental tissues (blood, brain regions, etc) to the chemist types you have to deal with control issues. You can't just analyze the raw blood (or even plasma) or throw a few mg of brain tissue into the machine you know. There will be some so-called extraction steps at the bench to clean up your samples, removing a bunch of other proteins, molecules and gunk that isn't of primary interest. Since it is in many cases uncertain what effects this may have on your quantification of, say, methamphetamine or dopamine or 5-HIAA, you have to include parallel control samples in which you have spiked the sample with a known quantity of what you are trying to measure. This allows you to correct for the extraction (and storage, and...) procedure in your test samples by reference to what changes were produced in the known sample. If you have multiple compounds you want to quantify? Well, depending on the situation, you may want to have multiple controls. Finally, in some cases you may want to spike the sample itself with a similar compound that will resolve as clearly analytically distinct so that you have the best possible control comparison.
And as many scientists can attest from their diverse experiences.....stuff happens. Analyses get blown, weird results are obtained, troubleshooting is required. In most scientific situations, we are not dealing with absolutely irreplaceable samples; after all the experiment can be re-run if necessary. In sports doping cases? Not so much. And therein lies the problem for any sports fans that have even a nodding acquaintance with analytical chemistry and the limitations detailed by Samia. I, for one, am under no delusions that the laboratory analyses of athletes' blood or urine samples are perfect and magical. So when there is any evidence of sloppiness, a lack of confirmation from multiple aliquots of the same sample, any evidence of contamination, etc....well, confidence in the verdict is not. very. high.