The use of saliva as an alternative to blood sampling for the assessment of the response to acute exercise or training was significantly stimulated by the correlation found between serum and saliva concentration of some hormones [19
]. However, other authors have warned against a straight analysis of metabolic changes in saliva based on the known effects of physical activity on blood metabolite levels [21
]. In addition, there is both a less pronounced effect on saliva metabolite concentrations compared to blood, and a shorter duration of responses [12
]. This fact is particularly relevant when assessing systemic variations caused by a physical activity, not homogeneous for all players, both in intensity and timing, like a soccer match. Moreover, the time period between the last significant physical effort of a given subject and the sample uptake is variable. This can explain in part, together with the expected interindividual differences, the high variation that the authors noticed in metabolite concentrations in the post-game samples, giving rise to high standard deviation values (41 against 90 for average SD in the pre- and post-exercise metabolite concentrations for the game group, respectively). These facts suggest that more useful data for assessing relevant aspects of an athlete’s performance, training status and health can be obtained if sampling times are more frequent, for example at halftime and immediately after substitutions take place. Despite all these limitations, our results show the potential of using our measurements to assess metabolic changes correlated with physical effort. For example, a PCA analysis of saliva showed a clear separation of pre- and post-samples, mainly driven by an increase in the concentration of most metabolites. Two interesting points emerge from this analysis. One is that the goalkeeper showed a very different variation, suggesting that metabolic changes sensed by saliva were sensitive to the type of activity and the players’ role. Secondly, increments in metabolite concentrations were not homogeneous, reflecting both a decrease in water content and different stimuli that the exercise puts at play.
This study also explored two different normalization schemes to account for the higher global concentration that is expected after exercise. A significant number of works report biomarker variations without taking into account any water-content effect that is provoked either by mouth breathing or the decrease in flow rate. In other cases, results have been expressed as a ratio to total protein concentration [22
]. However, this approach is only valid when no changes in the secretion rate of protein occur during or after exercise [23
]. Protein concentration increases, not only because there is a reduction in water content, but also due to sympathetic activity stimulated by physical exercise [24
], giving rise to an increment of certain types of proteins, not limited to amylases [11
]. For this reason, in this work, two different normalizations were explored, taking into account total protein (TPWS) and total observed metabolite concentration (TOMC). In the authors’ opinion, it is worthwhile to analyse the dataset using these two normalizations, together with the non-normalized values. In the latter case, valuable information can be obtained regarding the differential concentration effect that exercise provokes on different metabolites. Whereas, a comparison of the results between the two normalization schemes can afford insights into the different impact that exercise exerts on the levels of proteins and small molecules contained in saliva.
Furthermore, values of TPWS and TOMC are useful to classify the impact of the match on the salivary metabolome and to estimate dehydration. It was observed that the group that played the entire matched showed an increase of 40% in total protein concentration, against a 20% for the group that played only the last part. These values are in very good agreement with those observed after high exercise intensity (56%) and moderate exercise intensity (14%) [11
]. No increment in protein content was observed for the group that did not play at all or, interestingly, for the group that was replaced. This last result is in line with the fact that after a short period, saliva components return to their basal value.
A second way to estimate hydration status of the players is through the measurement of saliva osmolality and the flow rate [25
]. Although these data were not available in the present study, a variation in TOMC reflects a change in the total number of molecules in solution, and hence its variation is correlated to osmolality changes. In this respect, it is interesting to note that increase in TOMC for the game group after the match (around 80%) was much higher than that observed for TPWS (40%). This result implies that response of metabolites is greater than the raise in protein concentration, showing that water content is not the only factor influencing the change in saliva composition after exercise.
A third indicator of hydration is the urea:creatinine ratio [14
]. Values higher than 20 for the blood value was observed in the case of dehydration of hypoperfusion, mainly due to a high urea reabsorption causing a disproportionately elevated value relative to creatinine in serum. Both our measurements of this ratio in saliva, as well as other works using both NMR and MS to measure metabolite levels [15
], indicated that the basal value was approximately 2-6 in saliva, significantly below that observed in blood (10–20). The concentration of both urea nitrogen and creatinine are much lower in saliva than in blood. For example, salivary creatinine concentrations were 10–15% of those observed in serum, and were found to be unrelated in healthy subjects [26
]. Furthermore, exercise induced a significant decrease in the salivary ratio from 4 to 2 (p
= 0.024). In non-normalized data, this was caused by a significant increase in creatinine after the match, a variation that was not observed when normalizing by TOMC. Other studies presented creatinine levels in saliva before and after exercise, without taking into account the change in water content, and observed either a decrease [27
] or no statistically significant change [28
]. Our results of an increase creatinine concentration after exercise could be related to an augmented muscle breakdown due to acute tubular necrosis caused by exercise [29
Salivary lactate and correlated metabolites, like succinate and pyruvate, are essentially markers of the degree of tissue hypoxia reached during the exercise. Low-intensity exercises show constant lactate concentration since lactate production and removal occur with similar kinetics. On the other hand, when exercise exceeds the anaerobic threshold, a significant rise in blood lactate is normally observed [30
]. The correlation between blood and salivary lactate is controversial. The authors observed in an earlier study no correlation between hematic lactate and salivary lactate during a small-sided game session in male soccer players [32
]. Other studies reported higher correlation coefficients between lactate in blood and saliva during exercise [12
]. One important factor that determines the degree of correlation between the two lactate levels is the time left for filtration of lactate from blood to saliva [36
]. A significant increase in lactate only is found when non-normalized data are used, but not when the dataset is normalized by TPWS or TOMC. Regarding the total protein normalization, it was suggested that correlation between saliva and blood lactate was only noticed when the values were corrected by this factor [36
]. This study did not observe a significant increase in lactate when normalizing by proteins. However, it is conceivable that blood lactate was incremented after the soccer match. This lack of correlation can be in part explained by the different type of exercise, and more importantly, the variable time between the last high intense effort and the post-game sample uptake. Based on all these considerations, the change in salivary lactate obtained in these conditions may be only of limited value if it is intended to define training routines in order to enhance sport performance.
Two other energy related metabolites show an interesting opposite variation upon exercise. They are glucose and galactose. While the first showed a significant decrease, galactose levels were higher in the post-game samples, using both types of data normalization. It is tempting to speculate that this result is correlated to known differences between the two carbohydrates and exercise demand. For example, it was shown that the oxidation rate of orally ingested galactose was maximally 50% of the oxidation rate of a comparable amount of orally ingested glucose during 120 min of exercise [37
]. A different role of galactose and glucose in liver glycogen replenishment was also described, using drinks consumed during short-term post-exercise recovery [38
Metabolites like amino acids, urea and putrescine reflect physical effort on the body metabolism. Whereas the total venous plasma amino acid concentration after a 70 km cross-country ski race was observed to fall to approximately 60% of the pre-race level [16
], salivary amino acid concentrations were observed to slightly increase after physical exercise [39
]. It was observed that most amino acid levels were not significantly changed after the soccer match, with a few showing an increase, that is, glycine, histidine, phenylalanine, threonine and tyrosine. The elevation in aromatic amino acid levels is in line with that observed in blood [16
], suggesting an interesting correlation between blood and saliva for these metabolites. Other amino acids changing after exercise is taurine, for which a decrease in post-game samples was observed. Several studies have focused on the effects of taurine during physical activity [17
]. A diminished level of taurine was observed in rat skeletal muscles after exercise [40
], and in blood in a group of mice that behaved as spontaneous wheel runners [41
]. A similar decrease in salivary taurine levels was observed in young soccer players after a training session, and also correlated with distances covered by players during a game [32
]. The authors also observed an increase of putrescine in saliva, probably related to the known effect of polyamine accumulation in the skeletal muscle after physical exercise [42
]. It is thought that polyamine expression is important in aiding slow muscle fibers recovering from exhaustive exercise.
Hypoxanthine, a degradation product of ATP, was reduced in saliva after the soccer match. Purine concentrations were observed to be reduced in muscle and plasma, together with a decreased urinary excretion, after sprint training [43
]. This reduction in hypoxanthine levels in blood, muscle, urine and now saliva, likely represents a training-induced adaptation to minimize the loss of purines from skeletal muscle. This adaptation is advantageous in reducing the extent of replacement of the muscle nucleotide pool via the metabolically expensive purine de novo synthesis pathway.
Metabolomics studies, as the one here presented, offer the possibility of measuring the levels of a high number of metabolites in biofluids. In particular, when saliva is used to monitor the extent of metabolic response to physical exercise, having access to a panel of different biomarkers instead of just a small number, is of vital importance. When evaluating the systemic response to this type of stimulus, not only the competition itself determines the final effect, but also diet, hydration, training and global health of the athlete. The assessment of biomarkers should include select, diverse and well-validated markers of performance, health and recovery, which in turn can offer information regarding muscle status and oxygen transport, nutritional and hydration status, inflammation injury risk and muscle damage. From a practical point of view, however, it is necessary to obtain additional measurements sensitive to exercise intensity, such as the total distance covered, and when appropriate even respirometry, in order to normalize the observed metabolic changes with respect to the actual physical effort. On the other hand, the short time of the effect on saliva precludes analysis of the partial game group, and probably renders heterogeneous the measured changes provoked by the game. If used as biofluid, with all the recognized advantages, saliva needs to be taken shortly after the game, and in the case of players that are substituted, immediately after they finish their physical activity. Only a global vision of many biomarkers can help in understanding the complex reasons that govern metabolic response to exercise.