Incredible Biologic Passport

28 Jun 2012

"Limits and pitfalls of Athlete's Biological Passport" is the title of a recent article (Clin Chem Lab Med 2011, 49:1417-1421) by Giuseppe Banfi, expert on laboratory analysis, sports and doping, author of hundreds of scientific publications in prestigious journals of the international Scientific Community.

The Author criticizes the ABP method, discussing various topics that are not sufficiently clear or scientifically documented. In particular: 

- the Bayesian-like statistical model (ABP model) upon which the evaluation of the data is based is NOT freely accessible to the scientific community; 

- the source of reference data (eg. The mean value of Hb in cyclists) upon which the profile of the athlete is then evaluated is NOT clear enough

- the variance between subjects used for the cyclists is NOT correct: specifically, the (low) variance of Italian football players classified as endurance athletes has been applied to evaluate cyclists, while the (higher) variance a group of German athletes that included cyclists was not even considered. The use of an incorrect variance is crucial in the statistical evaluation of data; 

- the ABP method DOES NOT consider the seasonal variations in haematological parameters or the influence of training, de-training or altitude (although this is recommended in the very same Technical Document of the ABP method); 

- the profiles, shown in Excel format, DO NOT take into account the time difference between samples, displaying only in their succession, causing the observer to overestimate the fluctuations of the values. The interval between samples, > 5 days and < 3 months, as correctly suggested by the authors of the ABP, has NOT been observed. As the haematologic parameters change over time, especially in periods of intense training, the timing of the tests is crucial in a system based on variance; 

- the possibility of false positives in the ABP is NOT evaluated correctly. In fact, if an athlete is tested once, the probability of a false positive with respect to the confidence intervals will actually be 1/1000, whereas if he's tested 20 times and the probability is about 20 times higher. Just like the variance not being correct (too low), the probability of false positives will be higher; 

- despite the rules on collection, transportation and storage of blood samples being clearly described and recommended in the Technical Document of the ABP, in practice they often are NOT being followed. In particular, the reticulocyte count is stable only up to 10 hours after sampling even when the low temperature storage is certified. If these basic rules are not complied with the test results are NOT reliable

- the Quality Control of the instruments used for the analyses was NOT properly secured, nor has analytical variability been sufficiently considered in the ABP program; 

- missing "expected changes" in hematological parameters have often been considered when assessing the guilt of the athletes: particularly, if Hb does not drop during stage races the athlete is accused of illegal practices. In this respect, the published data are scarce and controversial because of a very high interindividual variability. 
In any case, the finding of CONSTANT values of a parameter can NOT be assessed by a method (ABP) based on VARIATION of these values. 

As a further invitation to reconsider the credibility of the ABP system, I'll add an episode of which I came to know. 

In the spring of 2010, some riders of a Pro-Tour team that were training at altitude (on Teide, Tenerife) were subjected to the normal ABP samples. One they received the results of the analysis, considering them unreliable (the values were too high), all it took was a phone call from the team doctor to his friend Dr. Zorzoli, in charge of UCI's doping department, in order to get the results of those tests cleared from the profiles, as deemed inconvenient for the Team and for the sake of the Biological Passport system, which tends not to consider the effects of altitude. 
This behavior reminds me of the rather common habit of certain Researchers to hide or simply ignore "inconvenient data", i.e. furthest from the "truth" that they want to prove. 

Very recently the UCI opened an ABP proceeding on a cyclist, considering data going back 3-4 year from now. 
Of course the "experts" who assessed the case were NOT aware of the name of athlete, as the regulation prescribes and as it probably happened in all the cases that have previously been sanctioned. 
I would say that a snowball in Hell has a better chance...