Basketball is an intermittent team sport, characterized by alternating phases of low and high intensity work during four 10-min quarters. Players combine explosive and powerful accelerations, jumps, sprints, and specific movements such as rebounding, jump shooting, lay-ups, closing out, shot blocking, high-speed play, and fast breaks (Vázquez-Guerrero et al., 2019). Several factors can affect the movement demands in basketball, with player position being the most important (Pion et al., 2018) because each position is associated with its own duties and therefore requires specific movements. For example, backcourt players (i.e., point guards and shooting guards) run more than frontcourt players (i.e. centers) (Scanlan et al., 2011; Scanlan et al., 2012; Scanlan et al., 2014a). In contrast, point guards run more compared to forwards and centers (Hůlka et al., 2013). These differences reflect the nature of the positions. While backcourt players are generally smaller, quicker, and responsible for ball transitions, which require more movements, frontcourt players tend to be bigger, slower, and have more physical contact as they play in a smaller area (Kucsa and Mačura, 2015; Ibáñez et al., 2018). Such data is a result of external load monitoring studies using different methods of analysis among which video-based time motion analysis (VBTMA) is the most popular. However, these studies are based on predetermined speed thresholds that could result in over/under-estimate external load since player’s running speed vary depending on their playing levels, position, physical fitness, genetic background and other factors. Therefore, individualization of thresholds would make VBTMA a more accurate, valid approach (Abt et al., 2020).

The difference in movement patterns between different player positions indicates various contribution from aerobic and anaerobic metabolic pathways in each position during basketball games (Ferioli et al., 2018). In a previous study, the metabolic load of different positions (2 guards, 3 forwards, and 3 centers) in five games was compared, showing that guards have higher heart rates (HR) than forwards or centers (Vaquera et al., 2008). However, it is likely that HR underestimates the metabolic load in basketball because of its intermittent nature (Lambert and Borresen 2010), meaning that better methods to study the metabolic load are required. In this context, metabolomics can be a good alternative since it is a comprehensive, non-invasive approach, which can identify high numbers of metabolites simultaneously (Pohjanen et al., 2007). To date, previous studies have used metabolomics to distinguish the metabolic profile of athletes from different disciplines (Al-Khelaifi et al., 2018). However, the only study using metabolomics to investigate team sports is a comparison between the quarters of a game in elite male basketball players (Khoramipour et al., 2020).

The association between movement patterns determined by VBTMA and metabolic profiles assessed by metabolomics would provide a further understanding of the interaction between player movement and the metabolic consequences. Currently, only three studies have explored the correlation between indicators of internal and external loads in basketball games. Of these, Scanlan et al. (2014b) found moderate correlations between accelerometer data and session ratings of perceived exertion (sRPE), training impulse (an index coming from a combination of heart rate during training and rest) and summated-heart-rate-zones (adding up the time spent in each of five HR zones) in semiprofessional male basketball players. Furthermore, Svilar et al. (2018) reported that player load, acceleration and deceleration, jumps, and changes of direction were highly correlated to sRPE. More detailed internal load analyses (e.g., using metabolomics) could provide a better understanding of the associations between external and internal load in basketball.

Therefore, the current pilot study aimed to compare metabolic profiles and movement patterns between different player positions during elite male basketball games. In addition, we explored the relationships between metabolic profiles and movement patterns as indicators of internal and external loads. We hypothesize that backcourt players have higher movement frequency, duration, and distance and show larger changes in aerobic metabolites, while metabolic changes in frontcourt players tend to be more anaerobic due to the nature of their positions, which require less dynamic but more static, strength and power movement patterns.

The present study shows that movement frequency, duration, and distance of high and low intensity running and -shuffling were higher in backcourt players, whereas static effort frequency and duration as well as jump frequency were higher in frontcourt players. Accordingly, specific indicators of aerobic metabolism such as taurine, succinic acid, citric acid, pyruvate, glycerol, acetoacetic acid, acetone, and hypoxanthine were higher in backcourt players, while indicators of anaerobic metabolism such as lactic acid, alanine, and 3-methylhistidine were higher and methionine was lower in frontcourt players. Furthermore, a regression analysis shows that acetylcholine, hypoxanthine, histidine, lactic acid and Leucine could significantly predict HIRW.

Logically following time motion analysis showing that backcourt players spend more time doing dynamic movements during the game, the coinciding metabolomic analysis indicates a greater reliance of aerobic metabolism compared to frontcourts. More specifically, we find higher pyruvate concentration and lower lactic acid in backcourt players which indicate that most of the produced pyruvate entered the Krebs cycle and was utilized in the aerobic metabolism (Mougios 2019; Schenk et al., 2021). A higher Krebs cycle activity is also supported by higher succinic acid and citric acid (i.e., Krebs cycle mediators) (Blackburn et al., 2020). A previous study showed that VO_2max was higher in backcourt than in frontcourt players (Abdelkrim et al., 2007), which is congruent with our data. Another study where %HR_peak was used to compare the metabolic load between different positions showed that guards’ (considered as backcourt players in our approach) mean HR corresponded to 88% of HR_peak (Hůlka et al., 2013), showing high reliance on aerobic pathways. However, HR cannot be considered a valid measure of basketball load monitoring since it may underestimate physiological load. Here, we instead assessed Krebs cycle mediators, which are aerobic metabolism indexes, directly (Mougios 2019) and therefore a more valid approach. Furthermore, higher levels of hypoxanthine, the end product of the purine cycle, further support a higher aerobic energy turnover and oxidative stress (Zhao et al., 2020) in backcourt players; and increased energy demand further led to increased purine cycle activity and subsequently higher hypoxanthine (Mougios 2019). In addition, taurine, the most abundant amino acid-like component in muscle and other organs, was higher in backcourt players (Huffman et al., 2014). We also found higher amounts of ketone bodies (e.g., acetoacetic acid and acetone) and glycerol as indexes of higher fat oxidation in backcourt players. As a practical consequence of the above findings of higher aerobic pathway activity and fat oxidation with more dynamic movements, a greater requirement for endurance training in backcourt players could be expected.

There are several explanations for the larger oxidative stress found in backcourt players. First, backcourt players perform more high and low intensity shuffling, which is considered to be the most intense movements in basketball (Abdelkrim et al., 2007) due to the consecutive changes of direction, acceleration, and deceleration (Abdelkrim et al., 2007; Scanlan et al., 2012). In this connection Schelling and Torres (2016) reported that guards (i.e., backcourts) show higher acceleration load mainly because of their lower body mass which allow them to accelerate rapidly. On the basis of the same reasoning, Reina et al. (2020) concluded that centers experienced lower overall load during the game. In contrast to our findings, Ferioli et al. (2020) reported a higher rate of high intensity movements as well as more time spent doing high intensity movements in centers (e.g., frontcourt players) than guards. Two main reasons could explain this controversy; they did not subdivide the various high intensity movements and included static efforts and jumps in this category. Here, our results show that the amount of static efforts and jumps are higher in centers. In addition, Ferioli et al. (2020) used qualitative methods for time motion analysis, in which intra-variability could be a source of error, in contrast to our more accurate quantitative analysis method.

Second, backcourt players’ roles are ball carrying and dribbling, since they are more skilled in ball handling (Scanlan et al., 2012) and in defending smaller players who perform more agile movements and changes of direction (Abdelkrim et al., 2007). This is supported by Ferioli et al. (2020) who showed that guards are the players with the highest ball processioning and when external load monitoring is conducted with ball possession, all movement categories are higher in guards. Backcourts are also the first players who start both offense and defense actions, resulting in more activity (Abdelkrim et al., 2007). In addition, previous studies using different approaches, such as sRPE, showed higher metabolic loads in smaller players (i.e. backcourt players) and especially point guards.

Third, backcourt players’ playing zone is out of the 3-point shot, which is less crowded and has more movement space, meaning that players can move more easily and thereby increase their activity. García et al. (2020) reported more covered distance in guards than centers who need to stay in close distance to the basket for rebounding. Playing in this small, crowed zone requires frontcourt players to have much body contact during rebound and positioning, which is evident by more static efforts (Abdelkrim et al., 2007; García et al., 2020) and leads to higher muscle damage clearly indicated by higher 3 mtyle histidine in frontcourt players. Accordingly, such intense physical activity is dependent on anaerobic rather than aerobic metabolism. This may also explain the higher alanine concentration in frontcourt players as alanine is the precursors of carnosine production, the main intracellular buffer.

In the current study, indicators of anaerobic metabolism such as lactic acid, alanine, and 3-methylhistidine were higher and methionine was lower in frontcourt players, indicating greater reliance of anaerobic metabolism. Previous studies have only measured lactate, in which higher lactate concentrations have been found in guards than forwards or centers (Rodriguez-Alonso et al., 2003; Abdelkrim et al., 2007). In contrast, Scanlan et al. (2011) did not observe such differences between positions, in which tactical approaches, player level, and playing style may explain the differences between these studies.

The only significant prediction of VBTMA data from metabolomics indicators found using regression analysis was for HIRW in backcourt players. This makes sense since HIRW is a demanding movement in basketball, which lead to the largest change in the metabolites (e.g., acetylcholine, hopoxanthine, histidine, lactic acid and leucine). Likely, the higher frequency in backcourt players lead to the significant correlation between HIRW and selected metabolites in these players. Although Fox et al. (2020) showed that player load (representing a measure of the accumulated load) was correlated with sRPE and session heart rate zone (sHRZ) in training sessions and games, our study was the first to correlate metabolomics as a measurement of the internal metabolic load, which provide more detailed insights than previous studies.

We had three main limitations in this study: First, we were not able to include the same diet for several days preceding the games. Since diet could affect the players’ metabolomes, our result should therefore be taken with some caution. Second, using the 500 MHz Bruker DRX HNMR, we may have missed some metabolic changes due to low power of the device. Finally, we explained the physiological basis for our findings, but the reader should be aware that subject properties (i.e. demographic and genetic), athletic experience, component of the team and game strategy could affect both internal and external loads, illustrating that our results should be interpreted with this in mind.

Based on novel metabolomics analyses, we demonstrate clearly different metabolic profiles of backcourt and frontcourt players during elite male basketball games; while backcourt players move more and show more aerobic metabolite changes, frontcourt players represent more anaerobic changes due to more static movement patterns. However, the relationships between these indicators of internal and external load were modest and the only significant prediction of movement category found was HIRW based on acetylecholine, hopoxanthine, histidine, lactic acid and Leucine in backcourt players.