The role of exercise in regulating bone health is still not well understood. However healthy bone is typically related to increased mechanical loading [1]. The magnitude of the strain could prevent and treat low bone mineral density (BMD) [2], [3] or increase the acquisition of bone mass during growth [4].

Cycling can be considered a healthy sport because it improves physical fitness and prevents fat accumulation [5], [6]. Adolescence is a sensitive phase for the acquisition of bone mass [7], [8]. Around 90% of bone mass is present at the end of the skeletal maturation phase [9]. Many professional and master cyclists can be classified as osteopenic [10], [11], [12]. Professional road cyclists have significantly lower bone mineral density (BMD) than the non-active population [10]. It is assumed that this non weight-bearing activity an insufficient stimulus to generate osteogenesis in clinically relevant bone sites [12], [13].

Rico et al. [14] did not find differences in the BMC in total or any regional site between adolescent cyclists and age-matched sedentary controls. Similarly, Duncan et al. [7] observed that female adolescent cyclists had similar BMD values for the whole body, lumbar, femoral neck, legs and arms than non-athlete controls. These researchers also found no differences among groups in mid-femur for the BMC and volumetric bone mineral density (vBMD) measured by computed tomography [15]. However, when these adolescent cyclists were compared with a group of runners, female cyclists showed significant lower values for BMD for the whole body, femoral neck and legs, and lower bone strength [15].

We hypothesized that cycling during adolescence is associated with lower bone mass acquisition compared a healthy adolescent population. Therefore the main aim of this investigation was to describe the bone status in adolescent cyclists compared to a healthy age-matched group. A secondary aim was to analyse the effect of years of cycling practice on the on the acquisition of bone mass.

Table 1 and Table 2 summarize the descriptive characteristics of the participants. Cyclists and control groups were comparable in age and height. Cyclists had significantly lower body mass, total lean mass, body fat percentage, total body fat and BMI (all, P<0.01). VO_2peak was higher in cyclists than in controls (P<0.01).

Raw data for bone related variables is presented in Table 3. In general, lower BMC, BMD and bone area was observed in cyclists compared to controls (P<0.05, Table 3). Bone values adjusted for the combined influence of height and total lean mass are presented in table 4. Compared to controls, cyclists had lower BMC for the whole-body, pelvis, femoral neck and legs (P<0.01). BMD at the legs, pelvis, total hip and in the trochanter and inter-trochanteric subregion was lower in cyclists than in controls (P<0.01). Cyclists had lower bone area than controls at the whole-body, legs and femoral neck site (P<0.01). Total hip area was greater in cyclists than in controls (P<0.01).

Table 5, 6 & 7 summarize raw data for bone based on whether cyclists and controls were younger or older than 17 years. Cyclists under 17 had lower BMC in the arms and legs and lower BMD in the whole body, pelvis, throchanter, arms and legs; and cyclists over 17 had lower BMC in the pelvis, femoral neck and legs, and lower BMD in the pelvis, hip, throchanter, femoral neck and legs (all P<0.05, Table 5, 6). Table 8, 9 & 10 present the bone values adjusted by height and total lean mass. When differences between groups were compared by age, cyclists under 17 had 10% lower BMC in the legs and 8% higher total hip area than age-matched controls (P<0.05). In groups over 17, cyclists had 26.5%, 15.8% and 14.4% lower BMC than controls at the pelvis, femoral neck and legs, respectively (P<0.05). BMD was lower in cyclists over 17 than in age-matched controls at the pelvis, hip, trochanter, inter-trochanter, femoral neck and legs (the percentages being 24.5, 10.5, 16.2, 7.5, 15.8 and 14.3, respectively, P<0.05). In addition, cyclists had a vBMD 17.1% lower than controls at the femoral neck (P<0.05, Table 9).

Analyses were repeated including fat mass as covariable with no variation of the presented results.

Group interaction by age was found for BMC at the pelvis, hip and inter-trochanter sub-regions, and for BMD at the pelvis, hip and femoral neck (all P≤0.05, Table 8 & 9), as well as for vBMD at the femoral neck (P≤0.001, Figure 1).

All the previous comparisons exhibited large effect sizes (Cohen's d ranged from 0.8 to 7), excepting the hip BMD which were low (Cohen's d = 0.3).

The main finding herein is that adolescent cyclists had lower BMC and BMD compared with healthy age-matched controls in regions of clinical interest (hip, pelvis and femoral neck). Our study also shows that differences in BMC and BMD between cyclists and controls were higher in adolescents over 17 years old.

The present study shows that adolescent cyclists had lower BMC and BMD than healthy age-matched controls. Cross-sectional and longitudinal studies have shown that weight-bearing and impact-loading sports improve bone mass, especially at weight-bearing sites [22], [23], [24]. However, athletes who perform activities in which the body weight load is diminished or without impacts, as in cycling, are associated with a lower bone mass compared with athletes who participated in weight-bearing sports [25], [26]. Male professional cyclists had lower BMD for the whole body (12%), legs (16%), pelvis (18%), femoral neck (25%) [22] and lumbar spine [28] than non-active controls of similar age.

The differences observed in BMC and BMD in our adolescent cyclists are similar to those observed previously in professional cyclists who trained 3 times as much [22]. Sanchis et al. [27] found that young tennis players had 69% of the inter-arm asymmetry in BMC observed in professional tennis players who trained nearly twice as much, and all the asymmetry in bone area. In the review literature, we have found only 3 studies evaluating the bone mass in adolescent cyclists [11], [14], [15]. Rico et al. [14] did not find differences in total or regional BMC between male adolescent cyclists with a similar training frequency than in the present study (10 h/week), and age-matched controls, when values were corrected by body weight; one possible explanation that may explain this discrepancy is that in our study we corrected the BMC and BMD by the total lean mass and the height, as they are the variables having the highest effect on bone growth [4]. Unfortunately Rico et al. [14] did not evaluate BMD.

Duncan at al. [11] observed that female cyclists had similar BMD at whole body, lumbar, femoral neck, legs and arms than non-active population, but lower BMD at whole body, femoral neck and legs than a group of female runners. The same researchers compared total and cortical vBMD at the femur bone in adolescent females from different sport disciplines (cyclists, triathletes, swimmers and runners) and a non-active control group of the same gender [15]. Duncan et al (2002) showed that BMC and vBMD in mid femur was similar in all groups, except for runners who showed higher BMD values and bone strength than cyclists [15]. Several aspects may cause the differences between this latter research and the present one, such as different control groups (sedentary vs. actives) [28], differences in lean mass [24], or the known gender dimorphism in the bone development [29].

Our results showed that differences in BMC and BMD between cyclists and active controls were greater in adolescents over 17 years old than in those under that age. We also found a negative association between age and BMC, and BMD, in the cyclists. Unfortunately we only can compare our results with longitudinal studies conducted in adults. Nichols et al. [29] described the tracked changes in BMD over a 7-year period in competitive male master cyclists and non-athletes. Their results showed that at the beginning of the study, cyclists had lower lumbar and hip BMD than the control group; interestingly at the end of the study master cyclists had lost more BMD than controls [32]. A previous study examined BMD over a one year season in amateur male cyclists and found 1–1.5% decrease in BMD at the proximal femur but no changes at the lumbar spine [33]. Nichols et al. [8] observed that master cyclists (>50 yr) had lower total, lumbar and hip BMD than younger cyclists (mean 31 yr).

Bone mineral density is the main variable used to determine osteoporosis [30]. There is a close relationship between BMD and bone mechanical strength [31]. Our study shows that adolescent cyclists developed lower BMD than controls at relevant clinical sites. This could increase the risk of bone fractures and/or osteoporosis. However, in spite of lower levels of BMD at clinical sites, adult cyclists develop a higher cortical thickness which can also increase bone strength [32].

A recent study showed that master professional cyclists (>50 yr) exhibited greater BMC and cortical area at the tibiae and radius which was associated with higher polar momentum of resistance [32]. Longitudinal studies should be conducted to corroborate this finding and to analyze whether this effect can be generalized to include other bones of greater clinical interest [33].

Some limitations should be recognized. One is the design, from which it cannot be concluded that the effect that is observed in older adolescents is due to the longer period (years) of practice of cycling rather than internal (i.e. genetic) or external (i.e. energy imbalance) factors. The absence of hormone and calcium intake data is another potential weakness of this study because this may affect bone acquisition; although it could explain the mechanisms behind these observations they should not change the found lower bone mass found in cyclists.

Nevertheless, the analysis of the interaction between bone mass, age and cycling training may indicate that the practice of cycling training is linked to the lower bone mass found in our adolescents. In the same line we have found a strong negative correlation, after taking into account the age, height, muscle mass and years of practice, between hours of practice and BMC and BMD in all the regions studied in older adolescent cyclists (r = −0.31 to r = −0.76) although none of them reached statistical significance, maybe because the low sample. Nonetheless, this hypothesis must be corroborated with longitudinal or intervention studies.

A strength of this study is that volumetric density was calculated: vBMD and BMAD have been proposed as a better reflection of the real bone density [20], [34]. We detect no other study where these bone parameters were estimated in adolescent cyclists. In our study, we found a 17% lower vBMD at the femoral neck of the older cyclists compared to older controls, which may be associated with an important risk of fracture in this relevant clinical zone. The BMAD in the cyclist group was 100% lower than that in the controls in the older adolescents; although of no statistical significance, maybe because of the sample size, this may imply important biological consequences reflected by the high effect size.