Key Points

  • Through longitudinal, individual and adaptive monitoring of blood biomarkers, the haematological module of the athlete biological passport (ABP) has become a valuable tool in anti-doping efforts

  • This literature review described the potential factors confounding the ABP to outline influencing factors altering haematological profiles acutely or chronically.

  • While our results support the current ABP paradigm as rather robust to delineate adaptive individual limits, our work may contribute to disentangling the numerous confounding factors of the ABP to gather the available scientific evidence.

Introduction

The concept of an athlete biological passport (ABP) was developed in the early 2000s and officially introduced in 2009 [1], marking a turning point in the anti-doping field. With the haematological and steroidal modules of the ABP, blood and urine variables were no longer utilised exclusively as direct screening parameters but as robust markers of either erythropoietic stimulation or steroid intake in individual, longitudinal monitoring [2]. This indirect screening approach is pertinent because indirect biological markers may reveal abnormal variations in blood and urine potentially induced by doping [3].

The ABP is implemented with an adaptive probabilistic model based on a Bayesian approach [4] to determine the probability that an athlete’s haematological variations are of prohibited origin [5]. Therefore, the effect of doping is not directly examined via the modification of the biological parameters but via the variation in the mean and standard deviation of these biomarkers [3]. When a first sample is collected, upper and lower thresholds are determined with population-based average benchmarks [5]. These individual limits are then subsequently and progressively adapted based on each athlete’s values as additional samples are taken [5].

Fourteen parameters are currently recorded and analysed as part of the ABP haematological module in the online Anti-Doping Administration and Management System (ADAMS), which was developed and supported by the World Anti-Doping Agency (WADA). In ADAMS, two primary markers are monitored with remarkably strict operating guidelines [1]: haemoglobin concentration ([Hb]) and the erythropoiesis stimulation index ‘OFF-Score’ (OFFs), calculated as OFFs = [Hb] − 60 × √Ret%, including the percentage of reticulocytes (Ret%) and [Hb] in g L−1. An atypical passport finding (ATPF) is generated when a primary marker value falls outside the athlete’s intra-individual range or when a longitudinal profile of primary marker values is outside expected ranges (sequence deviations), assuming a normal physiological condition’ [1]. A level of specificity of 99% (outliers correspond to the values outside the 99%-range i.e. at least 1:100 chance that this result is due to natural physiological origin) is required for the system to notify an ATPF.

The ABP uses this quantitative Bayesian model to detect an ATPF. Then, a qualitative expert review by the Athlete Passport Management Unit (APMU) handles administration of the individual passport before eventually requesting an evaluation by three independent ABP experts [4].

The latter expert review is essential because factors other than doping, such as physiological variations of biological origin or the results of an athlete’s activity [3], may explain variations in haematological biomarkers. Interestingly, the plasma volume (PV) is at the core of many haematological changes [6], and [Hb], which is, by definition, measured as a concentration, may significantly vary and become difficult to interpret if PV changes [7]. In an athletic context, red blood cells diluted in plasma will be affected differently by conditions such as acute and chronic exercise, environmental factors or certain illnesses [8], and these conditions can be defined as confounding factors leading to a potential misinterpretation of the ABP biomarker variations [9].

The variability caused by confounders can, however, be significantly reduced by understanding the nature of these factors [3]. Four main sources of variation may be outlined: pre-analytical and analytical conditions, physical exercise, environmental conditions and individual characteristics. Currently, very strict guidelines for blood collection and analyses [10, 11] address the effect of pre-analytical and analytical variations. In the current ABP operating guidelines [1], the notion of confounding factors is outlined only for the steroidal module (i.e. urine testing) of the ABP with reference to a recent review of confounders by Kuuranne et al. [12]. However, to the best of our knowledge, no systematic review has thoroughly examined the potential confounders affecting biomarkers of the haematological module. We hypothesise that the scientific evidence of confounders impacting blood profiles may challenge the current ABP approach. A review of the factors confounding hematological variables may thus help experts in their interpretation of abnormal ABP profiles.

Therefore, this systematic review investigates the existing literature regarding the various factors influencing haematological markers. Its results, presented in a narrative format, aim to enhance ABP experts’ prevailing understanding of these confounders.

Methodological Approach

Information Sources

Nine potential confounding factors of the haematological module were identified at the beginning of this review: doping practices, acute exercise, chronic training, exposure to a hot environment, exposure to a cold environment, exposure to a hypoxic environment, individual disorders or diseases, athlete characteristics and pre-analytical factors (Fig. 1).

Fig. 1
figure 1

Illustration of initially identified confounding factors

Full size image

A literature search was conducted in June 2020 on the PubMed and Google Scholar platforms with the above factors as research terms used in various combinations with the following terms: (haemoglobin OR haematocrit OR reticulocyte OR plasma volume) AND (haematology OR haematology OR haematological variation OR haematological parameter) AND (athlete) AND (sport).

Study Selection

From the results, the authors selected the most relevant articles considering the broad spectrum of different confounding factors potentially involved in blood variations. The main findings are presented in the result tables. Initially, all resulting titles and abstracts (n = 2325) were screened with adequate articles retained for further evaluation (n = 535). Grey literature was included by screening the references of the most relevant articles. Only studies in English with human subjects were considered for inclusion.

Eligibility Criteria

Studies where the effects of at least two confounding factors could not be clearly discriminated were excluded. Studies including both of the two primary ABP biomarkers (i.e. [Hb] and Ret%) were identified in a first selection round (n = 124). At this stage, however, no studies related to climatic conditions (heat and cold exposure) matched with the above-mentioned inclusion criteria, despite the fact that these conditions exhibit an impact on PV variations affecting ABP markers. Hence, several articles related to environmental conditions but reporting no direct measurements of Ret% or [Hb] were included in this review when PV changes (with a subsequent impact on [Hb]) were deemed pertinent. The process resulted in the selection of 82 pertinent studies in an anti-doping context where added value for the interpretation of ABP profiles was identified (as illustrated in the PRISMA flow diagram below in Fig. 2).

Fig. 2
figure 2

“PRISMA” flow diagram (Moher et al., 2009) Haemoglobin concentration ([Hb]); reticulocytes percentage (Ret%)

Full size image

Data Extraction

With the existence of strict WADA guidelines ruling out confounders mentioned in the doping control forms [1], this narrative review will, however, focus mainly on confounding effects not addressed by the latter WADA rules. By reviewing the multiple confounders identified (Fig. 1), some pre-analytical factors (e.g. acute exercise or exposure to various extreme environments) may exert a significant influence on the haematological biomarkers, increasing the risk of misinterpretation of blood profiles. The period of influence these factors ranges from a few minutes (e.g. body position during sampling [13]) to a few hours (e.g. intense exercise [14]), and can, therefore, alter blood variables. For this review, we have considered the relative changes observed after an intervention or in a specific condition for the following physiological variables: [Hb], Ret%, OFF-Score, haematocrit (Hct) and PV.

Results

The literature review indicates that doping practices represent a major confounder altering the ABP haematological variables most significantly (Tables 1 and 2). Other significant changes were observed after prolonged exercise (Tables 3 and 4), exercise training (Table 5), training periodisation (Table 6), thermal acclimation (Table 7) and hypoxic training (Table 8). Athlete characteristics (Tables 9 and 10) also produced significant changes. Less evident changes occurred due to pre-analytical conditions (Table 11) and acute exercise (Table 12).

Table 1 Changes of haematological variables related to rhEPO doping protocol
Full size table
Table 2 Changes of haematological variables related to various doping practices
Full size table
Table 3 Changes of haematological variables related to prolonged and multiday events in ambient conditions
Full size table
Table 4 Changes of haematological variables related to prolonged and multiday events in specific conditions
Full size table
Table 5 Changes of haematological variables related to various forms of chronic training
Full size table
Table 6 Changes in haematological variables related to training periodisation
Full size table
Table 7 Changes of haematological variables related to thermal acclimation protocols
Full size table
Table 8 Changes of haematological variables related to hypoxic training strategies
Full size table
Table 9 Changes of haematological variables related to athletes’ disorders or diseases
Full size table
Table 10 Changes in haematological variables related to athletes’ characteristics
Full size table
Table 11 Changes of haematological variables related to pre-analytical variations
Full size table
Table 12 Changes of haematological variables related to acute exercises in various environmental conditions
Full size table

First, differences observed in [Hb] related to doping practices ranged from + 10% following rhEPO doping [19] to − 14% after blood withdrawal [20]. Otherwise, the range of increase or decrease in [Hb] was − 5% due to certain pre-analytical conditions [79], + 16% after acute exercise [87], − 13% after prolonged exercise [30], − 8% after exercise training [41], − 2% after heat acclimation [51], + 9% after cold acclimation [57], + 4% after hypoxic training [60] and − 12% after blood donation [71].

Subsequently, a large amplitude was observed in the relative changes in Ret% with increases up to + 135% after rhEPO doping [15] and decreases of − 38% observed after blood transfusions [20]. The range of increase or decrease in Ret% notwithstanding doping was + 5% due to pre-analytical conditions [78], + 25% after acute exercise [83], + 63% after prolonged exercise [28], − 44% after apnea training [42], + 350% after hypoxic training [66] and + 27% after blood donation [70].

Differences in the OFF-Score as a result of doping practices ranged from + 18% after rhEPO doping [15] to − 38% after blood transfusion [20]. The margin of increase or decrease in the OFF-Score related to other factors was − 4% for pre-analytical conditions [80], − 15% after prolonged exercise [27], − 16% after exercise training [39] and + 22% after hypoxic training [58].

Hct differences ranged from + 10% after rhEPO intake [19] to − 15% after blood transfusion [20]. The range of Hct variations related to other confounding factors was + 4% [13] and − 4% [79] for pre-analytical conditions, + 12% after acute exercise [87], − 8% after prolonged exercise [33], − 7% after exercise training [41], − 2% after heat acclimation [52], + 5% after hypoxic training [58] and + 11% [72] and − 11% [71] after blood donation.

Finally, PV increased by 14% after chronic xenon inhalation [25]. The range of increase or decrease in PV linked to other parameters was + 4% for pre-analytical conditions [80], − 20% after acute exercise [87], + 24% after prolonged exercise [30], + 16% after exercise training [41], + 18% after heat acclimation [53] and − 14% after blood donation [72].

Discussion

Our study confirmed that the most obvious factor confounding the ABP biomarkers is blood doping. Doping protocols for erythropoiesis-stimulating substances (such as rhEPO injections) were generally structured with a treatment phase causing an increase in [Hb] and Ret% (ON-phase), followed by a reversal trend when treatment was stopped (OFF-phase) [15,16,17], although significant inter-individual variability was observed [16]. Following blood withdrawal there should be a decrease in [Hb] and increase in Ret%, with the opposite effect occurring after re-infusion [20, 21]. In addition, chronic exposure to low doses of carbon monoxide was recently shown to positively influence erythropoiesis and alter markers sensitive to PV variations [23, 94]. Conversely, repeated intake of desmopressin or chronic xenon inhalation induced haemodilution and decreased concentration-based biomarkers sensitive to PV shift [24, 25, 95]. However, micro-dosing doping schemes complicate the analysis of blood profiles due to limited haematological variations [22]. Indeed, the efficiency of the ABP appears significantly lower with low-volume autologous transfusion protocols due to several factors that may influence its sensitivity (e.g. timing of the sample collection) [22]. The current findings outline, for instance, the need for a strategy able to detect minor blood manipulations [96].

Numerous studies have highlighted specific blood variations as the result of pre-analytical factors including circadian modifications [73, 78] and prevention strategies such as sodium intake [79], overhydration [80] or posture adjustments [13]. In the same way, a temporary [Hb] increase caused by PV reduction was usually observed following acute exercise [7, 83, 85, 86] without affecting Ret% in most cases [84]. Similarly, acute exposure to extreme environmental conditions, such as heat [87,88,89,90], cold [91, 92] or hypoxia [93], was reported to cause transitory PV shifts. Nevertheless, the strict WADA guidelines incorporate various pre-analytical precautions (e.g. reporting any exposure to hypoxia or extreme environment and other pathological conditions) to account for possible pre-analytical variations. However, other factors are not considered in the current model and may alter a proper interpretation of ABP profiles.

Prolonged exertion may affect an athlete’s blood values with temporary but delayed effects persisting for several days [26], and these effects may occur and persist independent of environmental conditions. A progressive decrease of [Hb] was observed during multiday events (e.g. cycling stage races) [27, 29, 30, 36, 37], although other studies observed a stabilisation over time [31, 33] or even an increase towards the end of the competition [32]. The Ret% was mostly unaffected by participation in multiday events [27], although some variations (unrelated to an erythropoietic stimulation) were reported [26, 28].

In contrast to transient variations caused by acute exercise, PV was shown to expand over a few weeks of endurance training, thereby reducing [Hb] [39,40,41]. Interestingly, multiple studies also reported a noticeable variation in Ret% after a few weeks of training [39,40,41]. Furthermore, acute and chronic training loads, the competition calendar and training periodisation were shown to significantly alter blood variables included in the ABP [43, 48, 49, 97], with [Hb] decreases most likely during periods with the highest training loads [46]. In a recent 12-month longitudinal study of elite cyclists addressing the influence of training on ABP variables, no ATPF was observed, underlining the relative robustness of the ABP adaptive model in incorporating varying training loads [97].

Various environmental conditions (e.g. hypoxic or hot environments) are currently gaining popularity as additional training stimuli [98] with a putative effect on fluid balance [99]. PV increase represents one of the main physiological adaptations that occur during heat acclimation strategies [8, 100], and this increase is, in turn, often linked to [Hb] reduction [50,51,52,53,54]. Although PV varies after heat acclimatisation, such variation may also occur without influencing ABP values [55], depending on the type of acclimatisation strategy and subjects involved. Another form of exposure to extreme environments is dry sauna bathing [101], which resulted in some heat acclimation with a significant PV expansion (up to ~ 15%) [52, 53]. A recent systematic review minimises the confounding risk of heat acclimation because no significant change in PV variations was reported after various acclimation protocols [102, 103]. Conversely, several sessions of whole-body cryotherapy were proposed to decrease [Hb] [56].

Hypoxia is another environmental condition now widely used by athletes [104] where [Hb] may be either not [58, 59] or only slightly affected [60] depending on the timing of blood sampling after living high-training high (LHTH) periods. Moreover, an increase in Ret% was reported during hypoxic exposure [59, 61], while Ret% decreased upon return to a lower altitude [58]. Similar blood variations were reported following living high-training low (LHTL) protocols with, however, a more pronounced increase in [Hb] [62,63,64,65]. Variations due to hypoxic training observed immediately after altitude exposure were reported to persist three weeks after returning to sea level [58, 62]. In addition, intermittent hypoxic exposure (IHE) or training (IHT) may positively influence Ret% [66]. Nevertheless, the rationale for the use of IHE for erythropoietic purposes in athletes is limited [105], and the response is inconsistent when hypoxic exposure is not prolonged [67, 68]. Prolonged exposure to simulated or real altitude was reported to induce similar and prominent changes in total haemoglobin mass (Hbmass) [106, 107]. Nevertheless, haematological variations following hypoxic training are contradictory [59, 104] and not systematically reported [63, 108], possibly due to initial fitness, initial fatigue or iron status [36, 109]. It seems unlikely, then, that training in a hypoxic environment could lead to a misinterpretation of an athlete’s blood profiles [65] if duly reported in the doping control forms as requested by WADA.

Finally, specific individual characteristics were also shown to impact haematological components in athletes. Some haematological disorders were reported to alter [Hb], with these athletes exhibiting lower values than those of healthy athletes [69]. With some athletes suffering from haemochromatosis (a pathological condition requiring the withdrawal of large amounts of blood), [Hb] was decreased immediately after withdrawal [110] while a measurable increase in reticulocytes was sometimes delayed for a few days [70]. However, most haematological disorders observed in athletes were not associated with exceeding the limits of the model used to detect rhEPO [69]. Thus, the potential of these disorders to cause misinterpretation of the ABP profiles is limited. Finally, in contrast to Ret%, which is frequently higher among women [73, 111], [Hb] is known to be higher in men than in women. Ret% was also shown to vary during the menstrual cycles of active women with lower values reported in the follicular phase [74]. Still, most of these variations remained within the individual ABP limits. Furthermore, one should be aware that individual characteristics (e.g., sex or ethnicity) used to define initial individual limits (i.e. possible inter-individual variance) then become irrelevant since the adaptive model focuses on intra-individual variation for the correct interpretation of an athlete’s hematological profile.

Finally, while anti-doping blood samples are collected and analyzed following very strict guidelines [1], pre-analytical and analytical variations [112] should not be excluded from the studies included in this review since results for hematological variables originate from different analysers and/or varying pre-analytical procedures.

Conclusion

As this review shows, the effects of ABP confounders vary widely in amplitude and duration depending on the type of effector (i.e. doping, environmental condition, training or pre-analytical condition). Nevertheless, the absolute effects of the factors described in this review appear to be relatively limited when taken together. Furthermore, true and systematic effects of environmental conditions (i.e. heat acclimation or hypoxic training) on haematological biomarkers remain debatable due to significant differences in individual responses [103, 104]. Studies investigating specific ABP confounders generally report variations within individual limits of the adaptive model [13, 63, 65, 97]. The scientific level of ABP experts reviewing passports and their efforts to do so with the latter factors in mind can further limit misinterpretations of ABP profiles caused by the confounding factors identified in this review. Nevertheless, many authors have noted an important inter- and intra-variability concerning haematological biomarkers [111, 112], highlighting the need to improve the sensitivity of the ABP while interpreting confounders carefully. The present review contributes to a detailed understanding of confounding factors that could affect ABP biomarkers. By reporting limited variation in haematological variables through the selected studies of this review, our findings support the blood module of the ABP as an efficient instrument to deter and indirectly detect doping. Nevertheless, further studies on the confounders that affect blood variables may contribute to improving the module and thus fighting doping more effectively.