Quantification of testosterone‐dependent erythropoiesis during male puberty

What is the central question of this study? To what extent does testosterone influence haemoglobin formation during male puberty? What is the main finding and its importance? In boys, testosterone might be responsible for about 65% of the increase in haemoglobin mass during puberty. The underlying mechanisms are assumed to be twofold: (i) indirectly, mediated by the increase in lean body mass, and (ii) directly by immediate testosterone effects on erythropoiesis. Thereby, an increase in testosterone of 1 ng/ml is associated with an increase in haemoglobin mass of ∼65 g. These processes are likely to determine endurance performance in adulthood.

the individual development of Hbmass in adolescence. During male puberty, testosterone plays a crucial role in the growth of various tissues, and is reflected in a higher lean body mass (LBM) in men than in women. It is a well-known fact that testosterone also has a dosedependent stimulatory effect on erythropoiesis (Coviello, et al., 2008), although the underlying mechanisms are not yet fully understood.
During male puberty, the onset of testicular testosterone production is closely related to a subsequent increase in haemoglobin concentration ([Hb]), and the different testosterone concentrations are also assumed to be the main reason for the higher [Hb] in men than in women (Handelsman et al., 2018;Hero et al., 2005).
In general, [Hb] depends on Hbmass or red cell volume (RCV) and on the amount of plasma volume (PV), which are regulated by different hormonal mechanisms, i.e. mainly by erythropoietin and by the volume regulating hormones antidiuretic hormone, aldosterone and atrial naturetic peptide. In the short term, physiological changes in [Hb] are due exclusively to changes in PV; in the long term, changes in both Hbmass or RCV and PV must be considered (Prommer et al., 2018;Schmidt & Prommer, 2010).
During childhood, Hbmass, blood volume (BV) and PV similarly increase in both girls and boys during development, a change which is closely related to the increase in LBM (Prommer et al., 2018;Raes et al., 2006). Thereafter, [Hb] remains constant in girls and female adolescents until adulthood, while it increases in boys at the onset of puberty, a change which is closely related to the increasing level of blood testosterone (Hero et al., 2005). Because [Hb] also depends on the magnitude of PV it is not yet possible to quantify the absolute amount of haemoglobin or red cell mass which is produced under the influence of elevated testosterone levels during puberty.
In contrast to [Hb], Hbmass remains unaffected by changes in plasma volume and its measurement could quantify more precisely the degree of erythropoiesis during adolescence. Available data on Hbmass show a similar continuous increase in boys and girls until the age of ∼12 years, which in boys is followed by an exponential increase in the years afterward (Åstrand, 1952;Karlberg & Lind, 1955;Prommer et al., 2018;von Döbeln & Eriksson, 1972). In girls, however, just a continuous increase can be observed before and during puberty without any further change after that age.
To date, there exist no simultaneous measurements of Hbmass and serum testosterone concentrations in children and adolescents which could more precisely quantify the effects of increasing testosterone levels on erythropoiesis during male puberty. One reason for lacking data concerns health issues when applying the so-called gold standard method to children using radioactive tracer substances. With the optimized CO-rebreathing technique, there is now a simple and non-invasive method available to conduct such investigations (Gore et al., 2005;Schmidt & Prommer, 2005). The first aim of this study was, therefore, to determine simultaneously Hbmass and serum testosterone levels to estimate more precisely the relationship between testosterone and erythropoiesis in children and adolescents (Zachmann et al., 1974). Since both testosterone and Hbmass might be influenced by confounding environmental factors, a second aim was to

• What is the central question of this study?
To what extent does testosterone influence haemoglobin formation during male puberty?
• What is the main finding and its importance?
In boys, testosterone might be responsible for about 65% of the increase in haemoglobin mass during puberty. The underlying mechanisms are assumed to be twofold: (i) indirectly, mediated by the increase in lean body mass, and (ii) directly by immediate testosterone effects on erythropoiesis.
Thereby, an increase in testosterone of 1 ng/ml is associated with an increase in haemoglobin mass of ∼65 g. These processes are likely to determine endurance performance in adulthood.
investigate the possible relationship between training status or chronic altitude exposure with the pubertal testosterone concentration and with Hbmass.

Ethical approval
Ethical approval was granted by the ethics committee of the National University of Colombia at Bogota (reference: ID 06/2015). The study conformed to the standards set by the Declaration of Helsinki, except for registration in a database. Written informed consent was obtained from all children and their parents. The subjects volunteered to participate in the study and were free to withdraw at any time without any need to provide a reason.

Subjects
In total, 313 healthy and differentially trained children and adolescents (girls n = 94, boys n = 219; untrained group n = 123, trained group n = 190; training volume 15.0 ± 5.9 h/week) aged from 9 to 18 years participated in the study. All of the trained children and adolescents reported they practiced endurance sports, i.e. medium-and longdistance running, cycling, speed skating, race walking and triathlon.
The pre-condition for participation in the trained group was a training history of at least 2 years, a training volume of at least 6 h per week (see Table 1) and a training frequency of at least 3 times per week. The participants in the untrained group did not practice any sports, except school sports. The subjects were recruited in the Colombian regions TA B L E 1 Anthropometric data of the tests subjects classified according to their stage of maturation Training volume (h/week) Boys 11.5 ± 4.0 9.9 ± 3.7 14.5 ± 4.9 16.4 ± 5.7 19.9 ± 9.0 ≤0.000 Girls 11.9 ± 4.1 12.6 ± 1.1 15.7 ± 4.7 12.5 ± 3.3 19.0 ± 4.5 n.s. n.s.
Tanner, stage of maturation; LBM, lean body mass; training volume, hours/week in the trained group only. Effect sizes (Cohen's d) for the sex-specific change from the previous Tanner level are shown in italic next to the respective values; those for the differences between the sexes are indicated below the respective parameters. Significant differences from the previous stage of maturation: + P < 0.05, ++ P < 0.01, +++ P < 0.001; significant differences between girls and boys of identical stage of maturation: *P < 0.05, **P < 0.01, ***P < 0.001. No statistical analyses were performed for the parameter 'training volume' . The column on the right presents the results of the 2-way ANOVA with the between-subject's factors Tanner, sex, and interaction of both variables; n.s., not significant.
around Bogota (2600 m -3000 m above sea level, n = 163) and Cali (∼1000 m above sea level; n = 150). They had lived at their respective altitude for at least 5 years and had not changed their altitudes for more than 1 week during the past year. None of the females indicated the use of hormonal contraceptives and none used iron, folic acid or any supplements which could affect the parameters determined in this study. The anthropometric data of the test subjects are presented in Table 1.

Study design
The study was conducted in the laboratories of the Universidad Nacional de Colombia in Bogotá and of the Unidad Central del Valle in Tuluá, both in Colombia. After subjects' arrival, a medical examination was performed followed by an anthropometrical evaluation. Hbmass was determined using the optimized CO-rebreathing method. Two cubital venous blood samples, i.e. 4 ml heparinized blood and 8 ml for serum analyses, were taken for the determination of basic haematological parameters as well as testosterone, ferritin, erythropoietin and C-reactive protein.

Medical examination and anthropometrical evaluation
Health status was checked by medical examination including electrocardiogram under resting conditions. The status of biological maturation was evaluated by using the method of co-evaluation according to Tanner (1962), and then the children were classified into stages from I to V according to their external primary and secondary sexual characteristics.
For the anthropometrical evaluation, the five-component method was applied and data for body mass, height and body composition were obtained (Kerr, 1988;Stewart & Marfell-Jones, 2011). Skinfold caliper measurements were performed by the same scientist using triceps, subscapular, supra-iliac, abdominal, tight and medial calf.
Percentage body fat and absolute lean body mass were estimated using a correction for the age and sex of the participants (Slaughter et al., 1988). The typical error was 2.5% (Prommer et al., 2018).

Sample transport and storage
The whole procedure, including blood sampling, sample trans-

Haemoglobin mass determination
Haemoglobin mass was determined by using the optimized COrebreathing method as described by Schmidt and Prommer (2005) and modified by Prommer and Schmidt (2007). Briefly, a bolus of 99.97% carbon monoxide (CO; 0.7-1.2 ml CO per kg body mass) was administered to subjects and rebreathed along with 2-3 litres of 100% O 2 for 2 min. Because Hbmass might be influenced by training status (Prommer et al., 2018) the CO bolus was adjusted to a dose of 0.7 ml/kg (untrained) and 0.8 ml/kg (trained) for all children until the age of 14 years. Above 14 years old, girls received 0.8 ml/kg (untrained) and 1.0 ml/kg (trained), and boys 1.0 ml/kg untrained) and 1.2 ml/kg (trained). To compensate for the lower barometric pressure at altitude (ambient pressure at 2600 m approx. 600 mmHg), the doses were extended by the factor 1.25.
Arterialized capillary blood samples were taken from a hyperemized earlobe before and 7 min after the rebreathing procedure and analysed in sextuplicate using an OSM3 haemoximeter (Radiometer, Denmark). End tidal [CO] was assessed before and 2 min after the rebreathing procedure by a portable CO detector (Draeger Pac7000, Lübeck, Germany). After familiarization with the equipment and the breathing procedure, the children performed the method without any problems. Possible leakage, especially around the mouth and nose, was permanently controlled with a CO detector (see above). At the two laboratories, the identical equipment was used by the same staff. The typical error of this method obtained in our laboratory is 2.2% and is in accordance with the typical error published by (Gore et al., 2005).

Statistics
For statistical analysis the software SPSS Statistics version 25 was used (IBM Corp., Armonk, NY, USA). Data mostly refer to the stage of biological maturation and are presented as means and standard deviation for the respective stages according to Tanner (1962).
To compare the mean values of the different Tanner stage groups, a two-way ANOVA with two between-subjects factors, i.e., sex and training status, testosterone and EPO. As described above, and also before conducting these analyses, possible interactions between the independent variables were checked after drawing a directed acyclic graph. To estimate the regression coefficients for the relationship between normalized Hbmass (g/kg LBM) and these independent variables, a multiple linear regression analysis was performed.

RESULTS
The anthropometrical characteristics show the well-known time course during the development with increasing fat mass in the girls and a higher increase in LBM in the boys during puberty (Table 1).
[Hb] increased in the boys from 14.6 ± 0.8 g/dl at the age of 9 years to 16.0 ± 1.1 g/dl at the age of 18 years, while in the girls no difference was observed between the different age groups (Figure 1a). In contrast, Hbmass increased in both groups ( Figure 1b); it was, however, much more pronounced in the boys (from 365 ± 54 to 883 ± 136 g) than in the girls (from 297 ± 58 to 534 ± 128 g).
For the haematological data for the state of biological maturation ( Ferritin showed an opposite course in boys and girls; it was at the same level until stage III, then tended to increase in boys and to decrease in girls. Erythropoietin concentration was similar in both boys and girls and did not change during maturation (Table 2). Testosterone was mainly below the detection limit in both sexes up to stage II.
Thereafter, it increased significantly only in boys as far as stage IV and then tended to increase further at stage V (Table 2).

Regression analyses
When [Hb] and Hbmass of the male groups were related as dependent variables to serum testosterone as the independent variable ( Figure 2a, b), we found significant relationships for both analyses, which was closer for Hbmass (r = 0.761) than for [Hb] (r = 0.534;  Table 3.

DISCUSSION
The most important result from this study is the quantification of the relationship between the male-specific erythropoiesis and increasing serum testosterone levels during puberty. Before puberty, i.e. during Tanner stages I and II, Hbmass develops similarly in both girls and boys.
Thereafter, only a moderate increase occurs in girls to stage III, while in boys a steep augmentation occurs to stage IV. This results finally in approx. 65% higher Hbmass in post-pubescent boys than in the mature girls.

Testosterone
As

EPO
The serum EPO concentration showed no difference between the sexes and remained unchanged over the maturation process. This behaviour agrees well with data from Eckardt et al. (1990) and Yanamandra et al. (2019), who did not find any change in the EPO concentration in boys during puberty. The lack of influence of altitude and training status also agrees with the literature as neither of these exerts chronic stimulatory effects on EPO (Cristancho et al., 2016;Gunga et al., 2007;Jelkmann, 2011;Schmidt et al., 1999Schmidt et al., , 2002. As is discussed in detail below, it is assumed that the testosterone effects on erythropoiesis are caused in part by increased EPO stimulation (Bachman et al., 2014;Maggio et al., 2013  Effect sizes (Cohen's d) for the sex-specific change from the previous Tanner level are shown in italic next to the respective values; those for the differences between the sexes are indicated below the respective parameters. Significant differences from the previous stage of maturation: + P < 0.05, ++ P < 0.01, +++ P < 0.001; significant differences between girls and boys of identical stage of maturation: *P < 0.05, **P < 0.01, ***P < 0.001. The column on the right presents the results of the 2-way ANOVA with the between-subject's factors Tanner this time, serum testosterone increased by 4.9 ng/ml, which is similar to our results (+4.2 ng/ml). Similar effects are also observed in boys with delayed puberty receiving a testosterone treatment and increasing their [Hb] concentration by 1.6 g/dl (Hero et al., 2005). Krabbe et al. (1978) related [Hb] to serum testosterone level and found a good relationship of r = 0.7, which is closer than that demonstrated here (r = 0.54). This may be due to the delay of the testosterone-mediated effects on erythropoiesis which is assumed to be 4-5 months (Thomsen et al., 1986). The cross-sectional design of our study, therefore, may slightly reduce the relationship between testosterone and [Hb] during puberty.

Absolute haemoglobin mass
To date, Hbmass and/or red cell volume (RCV) has not been determined simultaneously with serum testosterone levels to evaluate the relationship between what is probably the most important influencing factor during puberty on erythropoiesis. In an approximation, Hero et al. (2005) calculated RCV using haematocrit and body mass and found a continuous increase by approx. 350 ml from the age of 11.7 years to 16.6 years in healthy boys. In delayed puberty, a similar increase was calculated after a 12-month treatment with testosterone (Hero et al., 2005). In our study, RCV rose during the same age (stage II to V), by 1140 ml, indicating that erythropoiesis was by far underestimated when RCV was calculated as mentioned above. Our data, therefore, show more accurately the real effect of puberty on erythropoiesis than changes in [Hb] and calculated RCV can. In our study, the difference in [Hb] between stages II and V was 11% (from 14.6 to 16.2 g/dl) while the difference in Hbmass accounted for ∼95% (from 428 to 832 g). Similar differences in [Hb] and Hbmass are observed between mature girls and boys (Tanner stage V) being approx. 15% and 65%, respectively. In other words, Hbmass and also the blood volumes of mature girls are only slightly above those values at the beginning of puberty.
The reason for the discrepancy in judging erythropoiesis by [Hb] or by Hbmass is the increasing plasma volume during growth and maturation. In girls, there occurs a parallel increase in Hbmass and PV resulting in an unchanged [Hb] during maturation, while in boys the percentage increase in Hbmass is considerably higher than the

Haemoglobin mass and testosterone
In both girls and boys a close relationship exists between Hbmass and LBM (Figure 2d). The augmentation of both parameters in girls during the whole maturation process and in boys to stage II occurs independently of testosterone and is probably regulated mainly by the (human growth hormone-insulin-like growth factor) hGH -IGF-1 axis (Vihervuori et al., 1996). One reason for the slowdown of erythropoiesis in girls during later maturation might be related to the appearance of female hormones generally completing the growth process and slightly counteracting erythropoiesis (Murphy, 2014;Peschle et al., 1973).
A close relationship of testosterone and Hbmass in boys first becomes obvious in stage III and, as can be observed in Figure 2b, there is a sensitive phase up to the plasma concentration of approx. 6 ng/ml.
When calculating the slope of the regression line for this concentration interval, the increase in testosterone by 1 ng/ml corresponds to an increase in Hbmass by ∼65 g. Above 6 ng/ml the efficiency slows down and higher concentrations are less related to erythropoiesis. A very similar behaviour was described previously by Hero et al. (2005) comparing testosterone and [Hb] and showing a similar narrow range for testosterone efficiency up to 5.7 ng/ml.
The erythropoietic stimulation via testosterone is likely to take place in two different ways. On the one hand, testosterone increases the LBM, which, as it was before puberty, is associated with a proportional increase in Hbmass. In addition, testosterone stimulates erythropoiesis in a special way that is independent of LBM, as becomes obvious from the normalized Hbmass (g/kg LBM), which increases from Tanner III resulting in a close relationship between normalized Hbmass and serum testosterone concentration (Table 3).
Several direct testosterone effects on erythropoiesis are discussed in the literature. Bachman et al. (2014) showed a significant increase in EPO concentration lasting several months during testosterone administration in the elderly, which, however, and according to our data, could often not be confirmed in healthy younger people (Bachman et al., 2014;Maggio et al., 2013;Shahani et al., 2009). Other possible effects are increasing bone marrow activity via androgenic receptors in erythroblasts (Claustres & Sultan, 1988) and testosterone-mediated increased iron incorporation into the erythrocyte precursor cells (Coviello et al., 2008) as well as changes in renal microcirculation (Murphy, 2014), which may stimulate the erythropoietic activity.
Interestingly from the blood compartments, only the Hbmass or RCV is directly influenced by testosterone. The PV, related to LBM, does not differ between boys and girls and shows almost no changes during the maturation process. Due to the constant PV and the increase in Hbmass/RCV in the boys, BV also slightly increases during puberty.
In addition, there are, however, some environmental factors, i.e. altitude and training status, which have affected Hbmass independently of testosterone in our study. As shown in Table 3, moderate altitude is associated with an increase in Hbmass by 1.1 g/kg LBM, corresponding to ∼8%, which is in accordance with data obtained from differentially trained male and female adults living at similar altitudes as the subjects in this study (Böning et al., 2004;Schmidt et al., 2002). To the best of our knowledge, these are the first data on altitude adaptation of Hbmass in children and adolescents, and show that adaptation to hypoxia also takes place at a young age. The higher training status is associated with an elevated Hbmass by 0.8 g/kg LBM (∼6%), which is almost identical to the values from Prommer et al.
(2018) for children between 9.9 and 12.4 years. It also shows that physical training in childhood and adolescence, similar to adulthood, has only a relatively minor influence on erythropoiesis (Schmidt & Prommer, 2008). In addition, it cannot be ruled out that the higher Hbmass in the trained group is possibly due to selection processes, as it represents a more advantageous basis for better endurance performance.

Practical impact
The data obtained here can also have a direct practical impact for clinical medicine and sports. Since the aerobic performance correlates closely with the Hbmass (a change of 1 g changes theV O 2 max by 4 ml/min) (Schmidt & Prommer, 2010), it becomes obvious that boys suffering from hypogonadism are massively limited in performance during daily life (Zitzmann & Nieschlag, 2003). Increasing the Hbmass through testosterone application will therefore very probably increase everyday performance. On the other hand, athletes with endogenously high testosterone concentrations in puberty should have an advantage.
According to Steiner et al. (2019), at the end of puberty there are already large individual differences in the levels of Hbmass among young athletes, which only slightly change in the following years.
Only those athletes presenting the highest Hbmass values at the age of 16 years are able to perform at highest national and international level during adulthood (Wehrlin & Steiner, 2021 Unfortunately, testosterone also plays a negative role in endurance sports. Before EPO was abused as a doping agent in almost all endurance disciplines, testosterone was used for the same purpose and probably still is (Vorona & Nieschlag, 2018). Interestingly, athletics is currently debating whether women with endogenously high testosterone levels should be excluded from international competitions (Handelsman et al., 2018). A maximum value of 5 nmol/l was set as the upper limit, which corresponds to a concentration of 1.4 ng/ml. Our data clearly show that exceeding this value would increase the Hbmass by at least 70 g corresponding to an increase inV O 2 max by at least 280 ml/min, leading to a remarkable advantage for the athlete.

Limitations
The correlations between testosterone and haemoglobin do not prove cause and effect, but only associations between the two variables.
Cause and effect can only be determined by testosterone application, which should not be done in healthy children due to ethical concerns.
Due to the change in [Hb] in hypogonadal children after testosterone treatment (Hero et al., 2005), however, it is very likely that general conclusions can also be drawn about the testosterone effect on the Hbmass level in children.
The LBM was not directly measured here, but estimated by skinfold measurements and, based on this, determination of the percentage of body fat took place. Although this method correlates sufficiently with the gold standard methods (dual-energy X-ray absorptiometry (Loftin et al., 2007), three component method (Aguirre et al., 2015)), the results should be interpreted with caution. Because of their ease of use and robustness, however, skinfold measurements have often been used successfully for the determination of fat mass and LBM in comparable groups of children and adolescents in Colombia (Spurr et al., 1992).

CONCLUSIONS
In conclusion, we have defined the relationship between testosterone and erythropoiesis in male puberty. Hbmass increases from the beginning of puberty until complete maturation by 33% in girls, and by 95% in boys. This is much greater than hitherto assumed from the increase in [Hb], which only amounts to 11% in boys. In the most sensible phase of puberty, the increase in testosterone plasma concentration by each 1 ng/ml is correlated with an increase in Open access funding enabled and organized by Projekt DEAL.

COMPETING INTERESTS
W.F.J.S. is a managing partner of the company 'Blood tec GmbH' , but he is unaware of any direct or indirect conflict of interest with the contents of this paper. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. were involved in the acquisition of data, analysis and interpretation of data for the work as well as in the critical revision of the manuscript.

AUTHOR CONTRIBUTIONS
All authors approved the final version of the manuscript and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.