Short-term effects of reduced energy availability on bone metabolism
A limited number of studies have investigated the short-term effects of energy deficiency (based on energy balance or EA) on bone metabolism in physically active individuals [
11,
51] (Papageorgiou et al., under review). An additional two studies have been conducted in normal-weight sedentary populations [
10,
21]. Ihle and Loucks [
10] explored the effects of three distinct levels of reduced EA, at 30, 20 and 10 kcal kgLBM
−1 day
−1, on bone metabolic markers in sedentary, eumenorrheic women, compared to a balanced EA at 45 kcal kgLBM
−1 day
−1. All three levels of reduced EA were achieved by a combination of dietary energy restriction (DEI: 45, 35 and 25 kcal kgLBM
−1 day
−1) and walking exercise (EEE: 15 kcal kgLBM
−1 day
−1) [
10]. This study showed a dose–response relationship between reduced EA and bone metabolic marker concentrations. Carboxyl-terminal propeptide of procollagen type 1 (P1CP) and total osteocalcin (both bone formation markers) were significantly reduced by 12 and 11% from baseline at 30 kcal kgLBM
−1 day
−1, whereas urinary amino-terminal cross-linked telopeptide of type I collagen (NTX) (a bone resorption marker) significantly increased (+34% from baseline levels) in the severely restricted EA condition at 10 kcal kgLBM
−1 day
−1 only [
10]. These data were, however, from sedentary women who differ from their physically active counterparts in bone characteristics (e.g., baseline levels of bone metabolic markers and bone strength), body composition and training adaptations [
4,
52,
53]. Additionally, the bone metabolic markers used to assess bone formation and resorption [
10] are not recommended as reference analytes by the International Osteoporosis Foundation and the International Federation of Clinical Chemistry and Laboratory Medicine [
20]. This is mainly due to the errors introduced when bone metabolic markers are determined in urine samples (e.g., requirements for creatinine corrections) and because of the heterogeneity of their fragments [
20]. For example, osteocalcin is found in different forms (i.e., carboxylated, under- or decarboxylated) and sizes (i.e., small, medium and large molecules), which may indicate bone formation, bone resorption or may even be more indicative of energy metabolism rather than bone metabolism [
54].
In order to examine the impact of reduced EA in physically active individuals, our recent study explored the short-term effects (5 days) of low EA at 15 kcal kgLBM
−1 day
−1, attained by diet (DEI: 30 kcal kgLBM
−1 day
−1) and exercise (EEE: 15 kcal kgLBM
−1 day
−1-running protocol) on bone metabolic markers in physically active women (Papageorgiou et al., under review). Low EA resulted in a significantly higher β-carboxyl-terminal cross-linked telopeptide of type I collagen (β-CTX) (reference marker of bone resorption) area under the curve (AUC) and a significantly lower amino-terminal propeptide of procollagen type 1 (P1NP) (reference marker of bone formation) AUC compared to a controlled balanced EA at 45 kcal kgLBM
−1 day
−1. The findings of this study confirm that altered bone metabolism, favouring resorption, following low EA is not confined to sedentary women [
10], but is also evident in their physically active counterparts. Some changes in hormones known to respond rapidly to states of energy deficiency, including reductions in insulin and leptin, were also reported in response to low EA, which is consistent with other short-term experiments [
10]. The actions of insulin and leptin on bone have been previously characterised and insulin receptors are present on both osteoblasts and osteoclasts [
55,
56]. In vivo, insulin increases bone formation [
57] and decreases bone resorption [
58]. Leptin may exert its bone effects directly through its receptors on osteoblasts and chondrocytes, but also, indirectly, by altering other hormones including oestrogen, cortisol, IGF-1 and PTH that may, in turn, mediate bone responses—for a review see [
59]. No changes in 17β-oestradiol concentrations (a marker of reproductive function) were reported in response to short-term reduced EA in physically active women (Papageorgiou et al., under review), which contradicts previous work in sedentary women that has shown reductions in oestrogen concentrations [
10] and reductions in luteinising hormone pulsatility [
31]. This could be due to the short timeframe and/or the solitary measurement of 17β-oestradiol levels in Papageorgiou et al. (under review) versus the 24-h pooled analysis conducted by Ihle and Loucks [
10].
Men may be more resistant to the negative effects of short-term low EA than women, due to factors such as the reduced energy cost of reproduction or a bone protective influence of androgens [
35]. In order to investigate this further, the same experimental protocol described above (Papageourgiou et al., under review) was conducted in a group of physically active men. In contrast to the findings reported in women, there were no significant alterations to either P1NP or β-CTX AUCs. High inter-individual variability in β-CTX and P1NP responses were, however, reported, indicating that low EA affected some, but not all men. Insulin, leptin or T
3 were not affected, supporting the overall absence of significant bone metabolic marker results. Zanker and Swaine [
11] demonstrated that energy restriction through exercise (60 min running) and a 50% restriction in estimated dietary energy requirements reduced P1NP levels in parallel with IGF-1 levels, but did not alter urinary NTX in trained men [
11].
In a preliminary, direct sex comparison of bone metabolic marker responses to low EA, no significant differences in bone metabolism were shown between active women and men; women—β-CTX: +19%, P1NP: −13%; men—β-CTX: +12%, P1NP: −14% (Papageorgiou et al., under review). Responses of hormones [insulin, leptin, T3, IGF-1 and glucagon-like peptide 2 (GLP-2)] to low EA did not vary between sexes, supporting the absence of sex differences in bone metabolic responses, which is in line with some [
60], but not all previous studies [
61,
62]. Low EA alters reproductive hormones (e.g., suppressed luteinising hormone pulsatility, reduced oestrogen levels) [
31] and contributes to the development of reproductive disorders in women [
32]. Future studies are needed to investigate the effects of low EA on markers of reproductive function, particularly in men, and their impact on bone metabolism. Studies to compare the contribution of markers of reproductive function to bone metabolism in response to low EA in men and women are also required, which will contribute to the RED-S paradigm, by integrating a direct sex comparison.
Long-term effects of reduced EA on bone health
No controlled, longitudinal or intervention-based studies have directly examined the response of bone to long-term reductions in EA. A number of cross-sectional studies are available that report data related to bone health and EA in track and field athletes [
63], endurance runners [
64,
65] and high-school varsity athletes competing in a variety of sports [
66]. These studies have primarily reported the frequency of Female Athlete Triad component occurrences and have concluded that 71 ± 27% of participants had reduced EA (<45 kcal kgLBM
−1 day
−1), 45 ± 16% had menstrual dysfunction and 34 ± 33% had low BMD (
Z score <−1). The aforementioned studies represent a wide range of athletes, and athletes from some sports may be more susceptible to the Triad components than others. For example, the prevalence study by Goodwin et al. [
64] was conducted using elite female Kenyan long and middle distance runners, and reported reduced EA (<45 kcal kgLBM
−1 day
−1) in 92% of the participants. In support of this, a systematic review investigating the individual and combined prevalence of the Female Athlete Triad in different groups reported that athletes competing in lean sports, i.e., “sports that place an emphasis on endurance training, low body weight, lean physique and aestheticism”, were more likely to exhibit one or more components of the Triad [
67]. The authors of this review did, however, report that variations in study design and methodological limitations, including the difficulty of measuring the Triad components in a field setting, limited interpretation of the results attained. It is important to consider that the studies described here [
63‐
67] were designed to assess prevalence of the Triad components and did not directly assess the relationship between low EA and bone health in active populations. Inferences related to the direct relationship between these components cannot, therefore be made.
Cross-sectional studies have provided some indirect insight into the long-term influence of energy deficiency on parameters of bone structure and function [
8,
9,
68‐
71]. Taken collectively, these studies support the hypotheses generated from the short-term studies described in “
Short-term effects of reduced energy availability on bone metabolism”, and indicate that bone metabolism and structure is negatively affected by prolonged exposure to an apparently insufficient DEI in relation to TEE in active individuals. De Souza et al. [
68] evaluated the independent and combined effect of oestrogen and energy deficiency [defined as resting energy expenditure (REE) ≤90% of predicted REE according to the Harris–Benedict equation] on bone metabolism in premenopausal exercising women. This study reported that energy deficiency had an independent and negative impact on bone metabolism, as evidenced by a suppression of osteocalcin in the energy-deficient group, while outcome variables related to REE were significant predictors of PINP and urinary CTX levels [
68]. More recently, the same group conducted a similar study to investigate the independent and combined influence of energy and oestrogen deficiency on bone, as assessed by peripheral quantitative computed tomography (pQCT), and showed that energy deficiency independently affected bone, with a main effect analysis reporting a relationship between energy status and tibial volumetric BMD, geometry and estimated bone strength [
69]. Zanker and Swaine [
71] examined the relationship between bone metabolic markers, indices of nutritional status and indicators of reproductive function in female distance runners. Evidence of down-regulated bone formation (reduced bone-specific alkaline phosphatase and osteocalcin) was prevalent among the group of energy-deficient individuals who were also experiencing menstrual irregularities (amenorrhea or oligomenorrhea) [
71]. Similarly, evidence from a series of studies on male horse-racing jockeys have shown that long-term exposure to reduced EA may negatively impact bone health in this group. Investigations have reported chronic energy restriction in an attempt to “make-weight” for racing [
72,
73] and EA below 30 kcal kgLBM
−1 day
−1 in male horse-racing jockeys [
74]. Additionally, reduced bone mass [
75], reduced cortical area and tibial strength strain index [
76] and increased urinary NTX/creatinine [
8] have been reported in jockeys, which were attributed to a chronic state of energy deficiency in this group [
77]. Wilson et al. [
70] reported lower BMD in male jockeys than in female jockeys, suggesting that the bones of male jockeys may be more susceptible to the negative effects of weight-making practices than their female counterparts. It is possible that the results reported within the study by Wilson et al. [
70] were due to a reduced frequency of weight-making practices in female jockeys. An alternative explanation, however, relates to a protective influence of adipose tissue mass on bone in females when compared to males [
78]. This may be mediated through the effect of oestrogen, which is involved in the regulation of bone homeostasis [
79] and is present in higher quantities in women compared to men. Adipose mass (which was reported to be higher in the female jockeys under investigation in the study by Wilson et al. [
70]) is a key source of aromatase, which contributes to oestrogen synthesis from androgen precursors [
80], and may have influenced the results attained [
70]. This hypothesis is speculative, however, and controlled studies investigating a potential sexual dimorphism in the response of bone to chronically reduced EA are warranted.
Reduced EA exerts a direct effect on bone [
68,
69]. It may also indirectly influence bone, through suppression of the hypothalamic–pituitary–ovarian axis, with subsequent reduction of oestrogen levels and associated reproductive dysfunction. In order to investigate this, a number of studies have been conducted on athletes experiencing menstrual dysfunction, which have shown reduced bone mass [
81‐
84], negative effects on bone microarchitecture [
3,
4,
85] and an increase in stress fracture incidence [
17] in amenorrheic or oligomenorrheic athletes, when compared to their eumenorrheic counterparts. These findings were largely attributed to a probable negative energy balance. Indeed, the negative bone effects reported in the energy-deficient groups in the studies by De Souza et al. and Southmayd et al. [
68,
69] were exacerbated in the group who were defined as being both energy and oestrogen deficient. Thus, Southmayd et al. [
69] suggested that the combination of energy and oestrogen deficiency is more detrimental to bone than energy deficiency alone.
Some studies have reported biochemical data in order to provide insight into the mechanistic pathways through which reduced EA may impact bone. Collectively, these findings provide evidence of energy conservation in apparently energy-deficient athletes [
8,
68]. For example, increased sex hormone-binding globulin, with a subsequent reduction in bioavailable testosterone, and decreased IGF-1 were reported as independent predictors of bone mass in a group of male horse-racing jockeys [
8], indicating a down-regulation of anabolic processes in this group. Similarly, De Souza et al. [
68] reported indirect biochemical evidence of energy conservation in a group of energy and oestrogen-deficient women, as evidenced by reduced T
3 and increased ghrelin levels. Kaufman et al. [
86] reported that resting metabolic rate was predictive of BMD in a group of elite professional ballet dancers. The authors suggested that the reported correlations between resting metabolic rate and BMD may be due to the nutritional habits of the dancers, a group who have previously been reported to routinely employ energy restricted diets, and to be prone to development of eating disorders, in an attempt to maintain the lean physique favoured in this aesthetic sport [
87]. Kaufman et al. [
86] suggested that the low BMD reported in their study may be caused, at least in part, by a suppression of resting metabolic rate in response to a negative energy balance. This suggestion is, however, speculative, since this cross-sectional study was not designed to directly investigate the relationship between energy balance and BMD in this group. Taken collectively the studies described within this section, support the theory that the body responds to reduced EA by down regulating basic physiological and anabolic processes. Further well-designed and controlled trials are required to confirm the mechanistic pathways through which this occurs.
An important factor to consider when evaluating the impact of reduced EA on bone health is the potential influence of sport type. Many of the studies described within this review were conducted using athletes who participate in sports that provide no-to-low impact and repetitive loading cycles, thus providing a low osteogenic stimulus [
9,
64,
65,
71,
88,
89]. Conversely, sports that convey high-impact, multi-directional movements and unaccustomed loads are thought to provide a higher osteogenic stimulus [
88] and there is some evidence to suggest that the mechanical loading afforded by particular sports may offer protection to bone under conditions of reduced EA. For example, participation in weight category sports may theoretically pose a risk to bone. Weight-category athletes generally strive to compete in the lowest weight class possible, as this is widely believed to convey a competitive advantage [
90]. This is often achieved by severely restricting DEI [
91], which may have consequences for bone, as described throughout this review. This notion was investigated by Dolan et al. [
9] who compared groups of age, gender and body mass index (BMI) matched weight-category athletes (elite amateur boxers and horse-racing jockeys) and reported higher total body, lumbar spine and femoral neck bone mass in boxers, compared to flat and national hunt jockeys. Regression analysis showed height and LBM to be the primary predictors of these differences. A group-specific factor was, however, also identified as influential and this was hypothesised to relate to the degree of mechanical loading experienced by both groups [
9]. This hypothesis is further supported by the results of Trutschnigg et al. [
92] who reported high bone mass in paper (46 kg) to middle weight (70 kg) female boxers when compared to active controls, despite a high prevalence of oligomenorrhea and low BMI in the female boxers. Similar results have been reported in gymnasts, a group frequently reported to have low EA [
93,
94] and high incidence of menstrual dysfunction [
50,
94,
95]. Gymnasts have also been reported to have higher BMD than non-gymnasts [
5,
6,
50], a finding which is often attributed to the high-impact nature of the sport. Caution must be taken when interpreting these results, since BMD is not necessarily a good indicator of bone strength in athletic populations, who may have higher strength requirements than the general population for whom the reference-based normative values have been developed. Indeed, a recent paper by Tenforde et al. [
96] reported that a high proportion of gymnasts were classified as “moderate–high risk” according to the Female Athlete Triad Cumulative Risk Assessment, and were subsequently more likely to sustain a stress fracture injury.
The available research indicates that bone is negatively affected by long-term exposure to reduced EA, although factors such as reproductive function and sport-type may mediate this response. It is important to note, however, that the majority of studies described within this section were not specifically designed to evaluate the influence of reduced EA on bone health, and, as such, the conclusions drawn from these investigations are indicative, rather than definitive. As identified in “
Energy balance vs. energy availability”, both DEI and energy expenditure (EEE or TEE) are difficult to measure accurately, and many of the available studies have used imperfect assessment modalities (e.g., self-reported food records). Chronic exposure to insufficient energy intake is often accompanied by macro- and micro-nutrient deficiencies, making it difficult to isolate the influence of reduced EA per se. The indirect evidence described within this section supports the hypothesis that the body may downregulate aspects of osteogenic function during times of reduced EA, in an attempt to conserve energy for more immediately essential functions. Well-designed and robustly controlled trials are required to confirm these theories and to more fully evaluate the mechanistic pathways through which chronic exposure to low EA may impact bone structure, function and metabolism.