Relating cardiorespiratory responses to work rate during incremental ramp exercise on treadmill in children and adolescents: sex and age differences

The results are reported as means ± standard deviation. Linear regression was used to evaluate the relationship between CPET slopes and body weight and between CPET slopes and \({ \dot{V}{\text{O}} }_{2,\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}\). The correlation between the variable was quantified with Pearson’s coefficient (r). The comparisons of the CPET variables and slopes obtained from female and male tests for different age groups were evaluated with a two-way ANOVA with Bonferroni for multiple comparisons (Origin software). The interaction effect between age and gender was investigated for the CPET slopes. Student t-test was used to compare regression lines between female and male groups and determine whether the parameters of the regression lines were significantly different. Difference of p < 0.05 was considered significant.

CPET variables for males and females measured among different age groups are reported in Table 2. Maximal HR was not affected by sex, whereas \({ \dot{V}{\text{O}} }_{2,\mathrm{G}\mathrm{E}\mathrm{T}}\), O₂ pulse, peak of \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\), \({\dot{\mathrm{V}}}_{\mathrm{e}}\), and \({ \dot{V}{\text{O}} }_{2}\) were higher in male than in female patients > 12 years of age. The peak \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) and O₂ pulse increased with age in both sexes.

The \(d{ \dot{V}{\text{O}} }_{2}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\), \(d{ \dot{V}{\text{O}} }_{2}/\mathrm{d}\mathrm{H}\mathrm{R}\), and \(d\mathrm{H}\mathrm{R}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) individual slopes were determined using data recorded before the gas exchange threshold, as reported in Fig. 2. The range of \(d{ \dot{V}{\text{O}} }_{2}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) slopes (Fig. 3a, 9.5–11 mL O₂ min⁻¹ W⁻¹) based on the estimate of \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}(t)\) across different age groups corresponded to a range of typical values of muscular efficiency from 30 to 26% (assuming an energy equivalent for oxygen of 20.9 J mlO₂⁻¹). The \(d{ \dot{V}{\text{O}} }_{2}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) slope was greater in male than in female patients > 12 years of age (Fig. 3a), and the effect of age was absent in both female and male groups. For both \(d{ \dot{V}{\text{O}} }_{2}/d\mathrm{H}\mathrm{R}\) and \(d\mathrm{H}\mathrm{R}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) slopes, there was a significant difference between females and males among different age groups (Fig. 3b, c): \(d{ \dot{V}{\text{O}} }_{2}/d\mathrm{H}\mathrm{R}\) was lower and \(d\mathrm{H}\mathrm{R}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) was higher in females than males. Furthermore, for both males and females \(d{ \dot{V}{\text{O}} }_{2}/d\mathrm{H}\mathrm{R}\) increased, whereas \(d\mathrm{H}\mathrm{R}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) decreased in older subjects. The interaction between sex and age was also statistically significant for both \(d{ \dot{V}{\text{O}} }_{2}/d\mathrm{H}\mathrm{R}\) and \(d\mathrm{H}\mathrm{R}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) slopes.

In Table 3 was reported the total work rate variation with time during the incremental ramp (\(d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}/dt\)). The \(d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}/dt\) slope increased with age in both females and males and was higher in males than females in groups older than 12 years old.

To further investigate the sex differences observed in \(d{ \dot{V}{\text{O}} }_{2}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\), its components \(d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}/dt\) and \(d{ \dot{V}{\text{O}} }_{2}/dt\) were analyzed separately. Specifically, \(d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}/dt\) (Fig. 4a) and \(d{ \dot{V}{\text{O}} }_{2}/dt\) (Fig. 4b) were plotted against body weight for both male and female groups. Both \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}/dt\) and \(d{ \dot{V}{\text{O}} }_{2}/dt\) have a strong linear relation to body weight. In males, the \(d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}/dt\) change with body weight was only 6% greater than that observed for females, whereas the \(d{ \dot{V}{\text{O}} }_{2}/dt\) change with body weight was 41% greater than that observed for females. Thus, the higher \(d{ \dot{V}{\text{O}} }_{2}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) slopes observed in males is mainly due to the higher \(d{ \dot{V}{\text{O}} }_{2}/dt\) because \(d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}/dt\) (i.e., workload changes) in females and males with the same body weight was similar (Fig. 4a).

To further evaluate the reliability of \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}(t)\) estimates, we tested whether any of the \(d{ \dot{V}{\text{O}} }_{2}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\), \(d{ \dot{V}{\text{O}} }_{2}/d\mathrm{H}\mathrm{R}\), and \(d\mathrm{H}\mathrm{R}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) slopes were related to body weight (Fig. 5) or \(\dot{V}{\text{O}}_{{2,{\text{~peak}}}}\) (Fig. 6). The \(d{ \dot{V}{\text{O}} }_{2}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) was not related to body weight (Fig. 5a) but it was significantly related to \(\dot{V}{\text{O}}_{{2,{\text{~peak}}}}\) (Fig. 6a). The \(d{ \dot{V}{\text{O}} }_{2}/d\mathrm{H}\mathrm{R}\) was linearly related to body weight (Fig. 5b) and \(\dot{V}{\text{O}}_{{2,{\text{~peak}}}}\) (Fig. 6b), whereas \(d\mathrm{H}\mathrm{R}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) was inversely related to body weight (Fig. 5c) and \(\dot{V}{\text{O}}_{{2,{\text{~peak}}}}\) (Fig. 6c).

The estimate of the peak \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) observed in this study increases with age as does \(\dot{V}{\text{O}}_{{2,{\text{~peak}}}}\) (Table 2). In this study, \(\dot{V}{\text{O}}_{{2,{\text{~peak}}}}\) and peak \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) obtained during incremental exercise on a treadmill are 10% and 25% higher than those obtained by (Cooper et al. 2014) incremental exercise on a cycle ergometer in populations with comparable age and body weight. Although these comparisons are limited by the fact that the cycle ergometer and treadmill data are not obtained from the same adolescent groups, the treadmill data are consistently higher than those of the cycle ergometer. The cycle ergometer involves less muscle mass recruitment when compared to the exercise on a treadmill (Krahenbuhl et al. 1985; Armstrong and Welsman 1994) which involves weight bearing muscle groups. In a study with children under the age of 10, \(\dot{V}{\text{O}}_{{2,{\text{~peak}}}}\) on a treadmill was 11% higher than that obtained on the cycle ergometer for both females and males (Turley et al. 1995). In another study on children, adolescents, and adults, a similar \(\dot{V}{\text{O}}_{{2,{\text{~peak}}}}\) difference was observed in female adolescents (Bar-Yoseph et al. 2019). Thus, a higher peak \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) for treadmill exercise than that for a cycle ergometer is consistent with the \(\dot{V}{\text{O}}_{{2,{\text{~peak}}}}\) differences observed between the two exercise modalities. The \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) expression allows for the detection of maximal \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) differences related to sex and age like those for \(\dot{V}{\text{O}}_{{2,{\text{~peak}}}}\).

The equation used to estimate \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) is based on biomechanical principles to study internal, external, and total work rates in human movement, (Cavagna et al. 1964; Cavagna and Kaneko 1977; Willems et al. 1995; Minetti et al. 2000) and used to optimize treadmill protocols (Porszasz et al. 2003). The external work rate quantifies the power required to move the body on a treadmill at specific speeds and inclines, but it does not include the work made by the joints to move the body’s center of mass (i.e., internal) (Herman 2008). Thus, we related the external to the total work rate for running based on the analysis of the kinetic and gravitational potential energy of the body’s center of mass (Cavagna et al. 1964; Cavagna and Kaneko 1977; Willems et al. 1995; Gosseye et al. 2010). Specifically, it is assumed that the fraction of the external work rate (\({f}_{\mathrm{e}\mathrm{x}\mathrm{t}}\)) is 70% of the total work rate during the incremental ramp protocol. The value of \({f}_{\mathrm{e}\mathrm{x}\mathrm{t}}\) is close to the range of values (60–75%) previously determined on a treadmill with a plate force to measure the force exerted under the foot during running (Schepens et al. 2001; Gosseye et al. 2010). A similar range was also observed in other studies (Minetti et al. 1994b, 2000; Willems et al. 1995; Saibene and Minetti 2003). Although the assumption of considering \({f}_{\mathrm{e}\mathrm{x}\mathrm{t}}\)=70% is consistent with the CPET slope characteristics, it might not fully reflect the individual body movement characteristics at different running speeds. In pediatric and adult studies on locomotion, \({f}_{\mathrm{e}\mathrm{x}\mathrm{t}}\) decreased from 75 to 70% (Minetti et al. 1994b; Schepens et al. 2001; Gosseye et al. 2010) for an increase in running speed in the range (50–85 m min⁻¹) used in our protocol. If these \({f}_{\mathrm{e}\mathrm{x}\mathrm{t}}\) changes with speed are considered, \(d{ \dot{V}{\text{O}} }_{2}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) slope decreases by 10% of the values obtained with a constant \({f}_{\mathrm{e}\mathrm{x}\mathrm{t}}\) (Figs. 2, 3). This potential effect on \(d{ \dot{V}{\text{O}} }_{2}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) is the same in female and male because the workload was the same for subjects of both sexes with the same body weight (Fig. 4a), thus sex differences observed in our work were not affected. Also, age-related changes in walking characteristics have been reported in children (Malloggi et al. 2019). Potential differences of body movement associated to age, sex, speed and inclination (Minetti et al. 1994a) are not considered in the expression proposed to quantify \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\). Nevertheless, the \(d{ \dot{V}{\text{O}} }_{2}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) (9.5–11 mL min⁻¹ W⁻¹) values obtained in our study are similar to those obtained for cycle ergometer exercise (Lai et al. 2012; Jones and Poole 2013; Cooper et al. 2014), thus indirectly supporting the validity of the proposed equation to estimate the total work rate. The \(d{ \dot{V}{\text{O}} }_{2}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) values observed among age groups correspond to the range of muscular work efficiency (30–26%) determined for cycling exercise (Whipp and Wasserman 1969). This evidence indicates that the \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) estimated for males and females at different age groups is consistent with the thermodynamic constraints of the biochemical processes in contracting muscle.

To further substantiate the approach proposed in calculating \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) and estimating CPET slopes at submaximal treadmill exercise, we verify that the relationship between CPET slopes and body weight or \({ \dot{V}{\text{O}} }_{2,\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}\) for the treadmill is similar to that observed for the cycle ergometer (Figs. 5 and 6). The inverse correlation between \(d\mathrm{H}\mathrm{R}/{d\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) and body weight and between \(d\mathrm{H}\mathrm{R}/{d\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) and \({ \dot{V}{\text{O}} }_{2,\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}\) are consistent with the linear relationship between \({d\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}/d\mathrm{H}\mathrm{R}\) and body weight and between \({d\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}/d\mathrm{H}\mathrm{R}\) and \({ \dot{V}{\text{O}} }_{2,\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}\) observed in cycle ergometer exercise (Cooper et al. 2014). Although \(d{ \dot{V}{\text{O}} }_{2}/{d\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) was not related to body weight, it was significantly related to \({ \dot{V}{\text{O}} }_{2,\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}\) (Fig. 5). Both findings echoed those reported for cycle ergometer exercise in the same age groups. Thus, the \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) estimated using the treadmill protocol is body size dependent and enables the analysis of size-dependent and size-independent properties of the CPET slopes similarly to the analysis obtained using cycle ergometer (Cooper et al. 2014).

The higher \(d\mathrm{H}\mathrm{R}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) and lower \(d{ \dot{V}{\text{O}} }_{2}/d\mathrm{H}\mathrm{R}\) slopes in females suggests a sex difference in the cardiovascular response to submaximal exercise (Fig. 3). Whereas the first slope quantifies \(\mathrm{H}\mathrm{R}\) change with increasing power output, the second slope quantifies the product of stroke volume and arterio-venous difference. Maximal cardiac output differences between females and males have been reported at different ages (Miyamura and Honda 1973). Specifically, in pediatric populations, males appear to have higher maximal cardiac outputs than females and this difference in part is related to a higher stroke volume in males (Rowland et al. 2000). Also, at submaximal exercise intensities, males relied on higher stroke volume with a lower HR (Turley and Wilmore 1997a, b; Armstrong et al. 2007) whereas arterio-venous oxygen difference was similar between genders. In these studies, the greater stroke volume observed in males was mainly attributed to a larger left ventricular muscle size (Turley and Wilmore 1997b). Thus, in our study, under the assumption that the arterio-venous oxygen difference in male is like that in female, the greater \(d{ \dot{V}{\text{O}} }_{2}/d\mathrm{H}\mathrm{R}\) in male patients was consistent with the finding of a higher stroke volume in males as compared to female patients. Furthermore, the higher \(d\mathrm{H}\mathrm{R}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) in females (Fig. 3c) was consistent with a higher amplitude of \(\mathrm{H}\mathrm{R}\) per \(\mathrm{W}\mathrm{R}\) observed at moderate and heavy exercise intensities in females (Lai et al. 2016). In another study relating CPET variables to work rate for treadmill exercise (Bar-Yoseph et al. 2019), the \(d\mathrm{H}\mathrm{R}/d\mathrm{W}\mathrm{R}\) slope in children and adolescents was similar to that found in our study. In this study, \(\mathrm{W}\mathrm{R}\) was determined imposing metabolic efficiency, whereas in our work, \(\mathrm{W}\mathrm{R}\) was independently derived from a biomechanical relationship between external and total work rate.

An age effect on the slopes was also observed in our study. The increase of \(d{ \dot{V}{\text{O}} }_{2}/d\mathrm{H}\mathrm{R}\) with age is in part related to the growth of left ventricular mass contributing to higher stroke volume (Daniels et al. 1995). It has been previously reported that the stroke volume of older children is higher than that of younger children for a given exercise intensity (Bar-Or 1983). Thus, the decrease of \(d\mathrm{H}\mathrm{R}/d{\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) with age observed in our study is consistent with a compensatory mechanism of an increase in stroke volume (indicated by an increase in the \(d{ \dot{V}{\text{O}} }_{2}/d\mathrm{H}\mathrm{R}\) slope). This cardiovascular adaptation in exercise is known in the pediatric population (Bar-Or 1983). Sex differences for \(d{ \dot{V}{\text{O}} }_{2}/d\mathrm{H}\mathrm{R}\) during growth as observed in our study in subjects older than 12 years old, can be partially explained by a greater growth in left ventricular mass in males when compared to females undergoing pubertal development (Pelà et al. 2016).

Our analysis was based on retrospective data obtained from exercise tests in which lean body mass was not systematically recorded. The analysis was mainly focused on corroborating the method proposed to estimate \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\). Although we cannot quantify the effects of body composition on CPET slopes, we can exclude any effect of BMI on sex differences because BMI was comparable in males and females at each age group (Cooper et al. 2016). To study sex differences in \({ \dot{V}{\text{O}} }_{2,\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}\) during development it is crucial to account for lean body mass (Armstrong and Welsman 2019b). Our work focused on CPET slopes, which in the case of \(dHR/d{WR}_{tot}\) and \(d{ \dot{V}{\text{O}} }_{2}/d\mathrm{H}\mathrm{R}\), should also be affected by body composition.

In conclusion, the results of this work indicate that the methodology proposed has the potential to be used as a tool in clinical and research settings to estimate the workload to determine the CPET slopes derived from an incremental ramp protocol using the treadmill. The estimates of the \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\) for treadmill were substantiated by verifying thermodynamic constraint and CPET slope characteristics. Nevertheless, future work should focus on further validating the proposed approach. A direct comparison of treadmill and cycle ergometer exercises as well as ways to change and measure work rate from individual locomotion characteristics for treadmill exercise can contribute to the design of validation strategies. As for the cycle ergometer, CPET slopes depending on \({\mathrm{W}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\), can be used to infer cardiovascular and metabolic function at submaximal exercise intensities in healthy and diseased states.