The Athlete Biological Passport (ABP)
The descriptions below use technical words and have been written for educational purposes only, it represents the view of the Swiss Laboratory for Doping Analyses and is not aimed to be used as a legal standard.
- What is an Athlete Biological Passport?
- What does an ABP contain?
- In which context has the ABP been introduced?
- How the ABP has been developed?
- Conceptually, how is the ABP used?
- Today, what is being biological traced in the ABP?
- What conditions must a marker fulfil to be part of the ABP?
- How to review an ABP?
- How are individual reference ranges established in the ABP?
- Does an abnormal outcome of the ABP means doping?
- Thus, what is done when the ABP shows an unusually large deviation?
- For the ABP, can a decision rule be defined on a true probability of doping?
- What is the logic of uncertainty in the ABP?
- What is the athlete's haematological passport?
- What is the athlete's steroidological passport?
- What is the athlete's endocrinological passport?
- What is a good test distribution plan for the ABP?
- What is the Athlete Biological Passport management tool?
- What services does the Swiss Laboratory for Doping Analyses propose for the ABP?
What is an Athlete Biological Passport?
The Athlete Biological Passport (ABP) is an individual electronic document that represents a collation of all data regarding a specific athlete that is useful in establishing whether that individual has doped. The fundamental principle of the ABP is based on the monitoring over time of selected biomarkers which can reveal either the effects of doping or a pathology. Because fair-play and athletes' health protection are fundamental in any anti-doping program, the benefits of adopting the ABP concept are far-reaching.
What does an ABP contain?
The preservation and tracking of information such as a longitudinal record of previous measurements of various markers of doping, age, sex, history of exposure to higher elevations, intake of over-the-counter and prescription medications, dates of participation in competitive events, etc... in an ABP provides a strong and accessible historical record by which to evaluate serial data and to recognize patterns in the marker concentrations that are characteristic of a pathology or doping. These elements are progressively entered in the course of the ABP development.
In which context has the ABP been introduced?
During more than four decades, anti-doping strategies have been quasi exclusively founded on the paradigm to discover a doping substance in a biological sample of the athlete. This paradigm has encountered sporadic success. The continuous development of new effective drugs resulting from today biotechnological race as well as the increasing sophistication of doping protocols have assigned limitations on this testing paradigm. Hence, in order to recover fair-play and protection of the athletes' health that this paradigm could not guarantee for example with the arrival at the beginning of the 1990's of synthetic erythropoietin produced by recombinant DNA technology, some sport authorities such as the Fédération Internationale de Ski and the Union Cycliste Internationale have introduced limits on markers of altered erythropoeisis with the athlete declared unfit and temporarily withdrawn from competing.
Markers of doping to testosterone have also a long history. For example, the testosterone over epitestosterone ratio is a marker of doping that has been used during many years to target athletes abusing from anabolic steroids.
Sometimes, the terminology “direct approach” - for the detection of the substance - versus “indirect approach” - with biomarkers stored in the ABP - is used to differentiate between these two paradigms. However, a clear and unambiguous distinction remains problematic in most cases.
How the ABP has been developed?
Conceptually, the ABP can be viewed as the mature result of various different products: the limits of indirect markers introduced in the 1990's by some sport authorities for anti-doping purposes, the longitudinal medical record introduced by the medical commission of some sports organisations at about the same period, the use of steroid profiling for targeting athletes to detect steroids abuse etc... These recent years, the introduction of multiparametric markers of doping, the use of the athlete’s own previous measurements to define individual limits of the markers with the athlete becoming his own reference, the inclusion of heterogeneous factors such as gender and age as well as potentially confounding effects such as exposure to altitude, the adoption of standardized protocols for sample collection and analysis, the use of external quality control systems to control analytical uncertainty, the development and validation of probabilistic inference techniques to evaluate the value of the evidence have all contributed to formalize originally disparate concepts into the ABP.
Conceptually, how is the ABP used?
The ABP represents a new paradigm in testing. Biological tracing throughout an athlete's sporting career is believed to be capable of application across all sports. In particular, the deterrent effect of the ABP has been high among the sports in which it has already been implemented.
Unusually large disparities between an athlete's historic values and values obtained from a recent test may alert officials of doping or a medical condition requiring closer examination. Both give good reasons to declare the athlete unfit and to withdraw him from competing during a short period of time, typically two weeks. Although this “competition rule” is not applied today by the sports authorities, there is a strong scientific consensus supporting its implementation based on the ABP.
When no pathology can explain large disparities observed in the ABP with doping remaining the sole possible cause, the information stored in an ABP is sufficient to launch a disciplinary, anti-doping procedure against the athlete.
Last but not least, an athlete can use his ABP to attest his fair-play via normal longitudinal profiles of biomarkers. While a negative outcome of an anti-doping (direct) test does not necessarily prove that the athlete is clean - because of the low sensitivity of some direct tests carried out at one unique moment in time -, the presentation of the ABP at the beginning of a competition can ensure that the athlete will participate in his natural, unaltered physiological state. If a competition rule is implemented, it would be impossible for the athlete to deviate significantly from his natural baseline levels so that the searched effects of doping would be too low in comparison to the risks taken.
Today, what is being biological traced in the ABP?
The ABP is constituted of several modules that are at various steps of progress. Today, the most matured module is the Athlete Haematological Passport (AHP). The AHP is a longitudinal record of indirect markers of altered erythropoeisis to identify the possibility of an athlete attempting to increase his oxygen carrying capacity. Nowadays, more than 800 elite riders have an AHP managed by the Union Cycliste Internationale. The Athlete Endocrinological Passport (AEP) is another module of the ABP that is based on a description of hormones secreted by the endocrine system. In particular, as subsidiary of the AEP, the Athlete Steroidological Passport (ASP) is composed of longitudinal steroid profiles to detect the abuse of testosterone or its prohormones. These three modules are detailed below. Additional markers coming from fields as disparate as proteomics, genomics, metabolomics, metabonomics, transcriptomics are studied nowadays and may be introduced in the ABP in the future. An advantage of the ABP is that the validation and introduction of a new marker in the ABP is for time immemorial. In contrary, for the detection of a doping substance by direct means, a specific test must be developed and validated for each drug arriving on the market. For example, there is no strong reason why the ABP would not be already sensitive to future generations of recombinant EPO, whereas the sensitivity of today “direct” test to future rEPOs is far from being guaranteed.
What conditions must a marker fulfil to be part of the ABP?
There are several conditions that a biomarker must fulfil to be included in the ABP.
Firstly, the measurements of the marker requires highly standardised procedures that follow justifiable protocols. In a medico-legal setting, it is the burden of the testing officials to demonstrate the validity and robustness of the measurements. This is particularly important for the ABP since it is essential to quantify the expected biological variations of the markers. Because the compliance to different protocols leads to different expected variations, these protocols are an integral part of the ABP. These last years, a great effort has been made to find a good compromise between a rigid standardisation that minimizes expected variations in the markers and prerequisites on the applicability of the ABP. For testing in biological specimens, there are protocols for the collection, transport and analysis of the samples.
Secondly, the marker must have a sensitivity to doping proven in longitudinal clinical trials. In particular, the relation between the sensitivity – the ability of the marker to detect doping in doped subjects (the percentage of true positives) – and the specificity – the ability of the marker to falsely detect doping in clean subjects (one minus the percentage of false positives) – must have been estimated on a large cohort of subjects. Empirical evidence on a high number of non-doped, control subjects is primordial since a high specificity – to avoid to falsely accuse an innocent – is required in an anti-doping setting.
Thirdly, the components of variations of the biomarker must be known in conditions when the protocols are followed. Some markers are known to be stable over time when measured on the same individual, that is to present a small within-subject variance. The monitoring over time of biomarkers is particularly of interest with the within-subject variations are significantly lower than the variations between subjects. The relevance of a longitudinal follow-up must have been assessed in longitudinal clinical trials by measuring the ratio of within-subject to between-subject variations.
Fourthly, in a widely heterogeneous population, the between-subject variability can be significantly reduced if one understands the nature of the factors that influence the marker. For example, higher values of the marker haemoglobin (HGB) are expected for males than for females. Therefore, the causal relationships between the heterogeneous factors - such as age, gender – and the marker must have been evaluated. Although factors such as ethnic origin and genotyping information are known to be related to the measured values of some biomarkers, they are not collected today in the ABP because of privacy issues. Confounding factors are on the other hand factors other than the investigated cause - doping or disease - that are known to have an influence on the marker. For example, transient exposure to altitude is known to modify markers of altered erythropoiesis. Knowledge of confounding factors that change from one test to the next typically lead to a decrease of the within-subject variance.
How to review an ABP?
The interpretation of the information stored in the ABP is then a typical problem of evaluation of scientific evidence under uncertainty. In clinical trials, volunteers receive a doping product and variations are observed in the biomarkers: doping – the cause - induces differences in the parameters – the effect. With the ABP, the goal is to establish whether an athlete is doped or presents a disease by examining the historical record of biomarkers measurements. In contrary to clinical trials, this type of problem goes against the causal direction and the only logical reasoning that may apply here is Bayesian reasoning. If an athlete receives a blood transfusion - the cause -, the value of the biomarker haemoglobin (HGB) increases - the effect. If a model that links cause and effect is available from empirical data obtained in a longitudinal clinical trial, Bayes’ theorem can be used to follow the direction that is opposite to that of causality and to determine whether an increase in HGB may be the result of a transfusion or is caused by natural variations. In particular, the causal relationship between a doping activity - the cause - and the induced modification in the biomarkers - the effect - can be formalized and graphically represented by a graphical probabilistic network called a Bayesian network. Bayesian networks allow anti-doping authorities to take into account firstly the natural variations of indirect markers – through a mathematical formalism based on probabilities -, and secondly the complexity due to the multiplicity of causes and confounding effects – through a distributed and flexible graphical representation. The strength of this approach is that it relies on sound empirical testing on large populations and justifiable protocols.
How are individual reference ranges established in the ABP?
An ABP comprises a series of tests collected from an athlete which enables individual limits for each biomarker to be established. As soon as new data is stored in the passport, Bayesian inference techniques are used adaptively to predict the likely profiles for future samples. In particular, as new test results are entered in the ABP, inference techniques allow progressively to switch the focus from comparison with a population to the determination of individual values. At each moment in time, for example in anticipation of a test carried out just before a competition, it is possible to predict expected values for all biomarkers in function of the information stored in the ABP.
In a medico-legal setting, it is necessary to ensure a very high specificity, because of the presumption of innocence and to avoid accusing falsely an innocent athlete. In a Bayesian decision scheme, this is equivalent to, firstly, assume a prior probability of doping equal to zero and, secondly, to set individual reference ranges for a high specificity. Typically, a specificity of 99% is chosen, with individual thresholds fixed at the 0.5 and 99.5 percentiles of the predictive distributions of expected values of the biomarkers for clean athletes. In this logic, any value lower than 0.5 or higher than 99.5 is considered as abnormal and will deserve closer scrutiny.
It is important to note that these individual reference ranges are known before the application of a new, successive test.
Does an abnormal outcome of the ABP means doping?
No, for two reasons.
Firstly, because the decision rule as described above is not based on a true probability of doping, but rather on “how the profile differs from what is expected in clean athletes”. This conceptual difference is well known in forensics for the evaluation of scientific evidence: to sentence an athlete solely from a high level of specificity would be a fallacy of statistical reasoning that results from misunderstanding the idea of multiple testing. A high number of anti-doping tests simply elevates the likelihood of finding a positive by pure chance alone.
Secondly, doping is not the only possible cause to explain a detected abnormality. A pathological condition must be excluded first. In haematology for example, the prevalence of blood disorders may be high in certain populations - typically a few percents - in function of factors such as age and ethnic origin.
That information in mind, it is not necessary to increase the specificity higher than 99% since the proportion of athletes presenting a medical condition may be significantly higher than 1% and can be remained undetected with a too permissive threshold. In comparison, in health-related fields, a reference range for a particular test or biological marker is usually defined as the values that 95% - or 2 standard deviations - fall into.
Thus, what is done when the ABP shows an unusually large deviation?
The ABP is reviewed by a panel of experts to determine the cause of the abnormality. This reviewing can typically be carried out during the short withdrawal of the athlete if a competition rule has been implemented. The panel of experts is composed of specialists in the field: haematologists for the evaluation of markers stored in the AHP, endocrinologists for the evaluation of markers stored in the AEP. The role of this panel of experts is not only to protect the athlete’s right to a qualified review prior to the possible assertion of an anti-doping rule violation, but it also ensures that all possible factors, causes and events are considered thoroughly.
For the ABP, can a decision rule be defined on a true probability of doping?
Yes, but only in conditions when the prevalence of doping can be precisely estimated. In the ABP, the appropriate rule for decision making is Bayes' rule. Bayes' rule is a rule for updating degrees of belief on receiving new evidence, typically new test results in anti-doping. In the process of evidence acquisition, Bayes' theorem is used to update a “prior probability of doping” to a “posterior probability of doping”. An accurate estimate of the prevalence of doping – that is the proportion of athletes who doped in a well-defined population – can typically be used for the prior probability of doping.
It may be thought that only a test able to easily identify drug cheats can be used to estimate the prevalence of doping. Let us present a simple example to prove the contrary. Suppose that 200 riders have been tested before the departure of a Grand Tour. The hematocrit of a population composed of Caucasian male endurance elite athletes living at low altitude is well described by a normal distribution, with a mean of 44% and standard deviation of 2.7%. Therefore, for a population with a zero prevalence, less than 4% of these athletes should present a hematocrit naturally higher than 49%. Suppose then that the test results returned 152 hematocrit values superior to 49%. Since 48 athletes presented a hematocrit below 49% in that population, only about 2 athletes should have had a hematocrit naturally higher than 49%. A rough estimate of the prevalence is therefore (152-2)/200=75%. This is a simplistic (actually biased) example and much powerful methods have been described in the literature to estimate the prevalence accurately.
What is the logic of uncertainty in the ABP?
The paradigm shift at work today in anti-doping follows the same lines of reasoning as the paradigm shift seen in forensic sciences in general, from archaic assumptions of absolute certainty and perfection to a more defensible empirical and probabilistic foundation. One of the main role of forensic scientists is to advise decision makers of the significance of their findings through the assessment of uncertainties associated with the inferences that may be drawn from evidentiary values. For this task, many forensic scientists consider probability theory and Bayes' rule as the fundamental concepts to govern their reasoning. In contrary to what is sometimes thought at first sight, forensic scientists do not see the numbers in probabilistic reasoning as being important by themselves, what is important is that a probabilistic framework provides sound rules of reasoning to check the logical consequences of some propositions. In particular, for the ABP, graphical probabilistic networks have the nice properties to firstly model the natural variations of markers of doping – through a mathematical formalism based on probabilities -, and secondly the complexity due to the multiplicity of causes and confounding effects – through a distributed and flexible graphical representation.
What is the athlete's haematological passport?
The athlete's haematological passport (AHP) is the module of the ABP that collects information on markers of erythropoiesis measured in blood samples. The AHP has the sensitivity to identify among other blood doping methods, recombinant erythropoietin abuse and any form of blood transfusion or manipulation. In the Anti-Doping Administration Management System (ADAMS) of the World Anti-Doping Agency, the following parameters resulting from a full blood count are today collected for the AHP:
- HCT: hematocrit;
- HGB: hemoglobin;
- RBC: red blood cells count;
- RET%: the percentage of reticulocyte;
- RET#: reticulocytes count;
- MCV: mean corpuscular volume;
- MCH: mean corpuscular hemoglobin;
- MCHC: mean corpuscular hemoglobin concentration.
In addition, the multiparametric markers OFF-score (index of stimulation) and ABPS (Abnormal Blood Profile Score) are calculated from this set of parameters.
Although all parameters are identified after the analysis of a blood sample, only the markers HGB and OFF-score fulfil today the conditions to potentially sanction an athlete. The other markers are used by the panel of independent experts as additional evidence to distinguish between blood doping, altered quality of the blood sample (eg. hemolysis) and/or the identification of a possible pathological condition.
There are six heterogeneous and confounding factors for the AHP:
- gender (fixed factor)
- ethnic origin (fixed factor)
- age (fixed factor)
- altitude (time-varying factor)
- type of sport (fixed factor)
- instrument related technology (time-varying factor).
The AHP is the sole module of the ABP that is used routinely today.
What is the athlete's steroidological passport?
The athlete's steroidological passport (ASP) is the module of the ABP that collects information on markers of an altered metabolism of endogenous steroids measured in urine samples. The ASP has the sensitivity to identify doping with testosterone and its precursors, drugs that act as oestrogen receptor antagonists and aromatase inhibitors.
There are 6 core parameters in the ASP:
- T: testosterone
- EpiT: epitestosterone
- A: androsterone
- E: etiocholanolone
- 5á-diol: 5 á-androstanediol
- 5â-diol: 5 â-androstanediol
- DHT: dihydrotestosterone
- DHEA: dehydroepiandrosterone
can be added.
The associated ratios T/EpiT, A/E, 5á-diol/5â-diol, A/T are calculated from these concentrations. At the exception of the T/EpiT that has been used during many years to screen for athletes abusing of testosterone, the ASP is today at the stage of a pilot project.
What is the athlete's endocrinological passport?
Apart from a steroid profile in urine, the athlete's endocrinological passport also includes information on markers of growth hormone abuse measured in blood. Markers such as insulin-like growth factor 1 (IGF-1), type-3 pro-collagen (P-III-P), insulin-like growth factor binding protein 2 IGFBP-2), insulin-like growth factor binding protein 3 (IGFBP-3) and carboxyterminal cross-linked telopeptide of type I collagen (ICTP) have all shown to be sensitive to growth hormone doping in longitudinal clinical trials. This work is still at the stage of development in the network of WADA accredited laboratories.
What is a good test distribution plan for the ABP?
An ABP program makes sense only if it is piloted by knowledgeable and qualified persons. The crucial question to make the deterrent effect of the ABP optimal is the timing of the tests. For the ABP, it is much preferable to have 4-5 tests performed every year from a well-thought, intelligent test distribution plan, than more results from tests scheduled randomly that will waste resources and dilute relevant information. Data already available in the ABP, coupled with the applicable competition calendar of the athlete and other non-analytical information must feed into the test distribution planning. Which biological sample (blood, serum, plasma and/or urine), the number of samples (1 or 2), which additional analysis to request to the laboratories (full blood count, CERA in blood or serum, rEPO in urine, growth hormone in serum, isotope ratio mass spectrometry in urine, etc...) are additional questions that require a qualified personnel to optimize the investments and resources deployed for the ABP and the anti-doping fight in general.
What is the Athlete Biological Passport management tool?
At the Swiss Laboratory for Doping Analyses, we have developed a software to store, manage, visualize and evaluate an ABP. The passport is an electronic document with extension “.abp”. The software is continuously updated in function of scientific advances in the discovery and validation of markers of doping. It only uses validated models published in the scientific literature. The software is only available for people or organizations working in the anti-doping field. Please contact email@example.com for any question about the ABP software.
What services does the Swiss Laboratory for Doping Analyses propose for the ABP?
In addition to the measurements of the markers in urine and blood samples, the Swiss Laboratory for Doping Analyses, through its Athlete Passport Management Unit, can assist anti-doping organizations:
- to implement the passport program in accordance with the WADA passport guidelines;
- to store and manage passports via the ABP software on an anonymous basis, with test results obtained from either ADAMS or an equivalent database;
- to provide a qualified evaluation of the passport that the organization can use for intelligent testing or other purposes;
- to liaise with an independent commission of experts chosen by the organization for the evaluation of passports.
Please contact firstname.lastname@example.org for any information on these services.