Effects of Low Energy Availability on Bone Health in Endurance Athletes and High-Impact Exercise as A Potential Countermeasure: A Narrative Review

Endurance exercise is energetically demanding, and it is recommended that athletes replenish their energy stores in a timely fashion to enhance recovery and avoid prolonged periods of low energy availability (LEA) [1, 2]. Energy availability is dietary energy intake minus exercise energy expenditure normalised to lean body mass or fat-free mass and is considered low when there is insufficient energy remaining to support optimum physiological function¹ [3]. During periods in which planned weight loss is a priority (when working towards “race weight”) it can seem counterintuitive to restore energy availability and athletes with rigorous competitive schedules may accumulate many transient periods of LEA. Furthermore, the time demand of high volumes of endurance training can often make it difficult for athletes to consistently meet their energy demands by limiting their opportunities to consume large meals, resulting in inadvertent LEA [4]. As a result, endurance athletes are at high risk of LEA with approximately 31% of female distance runners, and 25% of males, reported to suffer from this condition during training [3, 5].

It is proposed, in the Relative Energy Deficiency in Sport (RED-S) and Female Athlete Triad models, that LEA compromises bone health in athletes, including endurance athletes [1, 4]. LEA is characterized by the perturbation of several hormones involved in the regulation of bone (re)modelling, including the suppression of estrogen, testosterone, leptin, triiodothyronine, and insulin-like growth factor 1, and an increase in adiponectin [6‐10]. During a single year, 3–21% of endurance runners may suffer at least one bone stress injury (stress fracture and/or stress reaction) [11‐14]. Male and female endurance athletes with greater “Triad cumulative risk scores” are more likely to have suffered a bone stress injury [15, 16]. LEA also causes menstrual disturbances [17], the most severe being functional hypothalamic amenorrhea (FHA), a reversible cause of ovarian disruption characterized by the absence of menses and chronic estrogen deficiency [4]. The highest prevalence of menstrual disturbance is observed in sports emphasizing leanness and endurance sports such as running [18], where it may be as high as 50–65% [19, 20]. Female athletes with FHA suffer more bone stress injuries, and more miss training days due to injury, compared to their eumenorrheic counterparts [5, 21, 22]. Research suggests that high-impact exercises can be osteogenic despite low energy costs [23, 24]; however, their use as a tool to protect bone health and reduce bone stress injuries in athletes at risk of LEA has not yet been considered. In an environment where restoring energy availability (via an increase in energy intake or a reduction in training volume) may be difficult or not considered a priority, such a tool would be beneficial. Therefore, this review aims to (1) investigate the bone changes associated with LEA and (2) explore whether high-impact exercise could mitigate the bone changes associated with LEA. Given that FHA presents as a symptomatic marker of LEA in female athletes, and that FHA is highly prevalent in runners, the focus of this paper will be on weight bearing endurance athletes (WBEA) with FHA. Where the literature exists, evidence in male WBEA with LEA will be also be discussed.

Any exercise intervention in trained athletes, suffering from LEA, would require a unique set of characteristics to effectively protect bone from adverse effects. It is crucial that the intervention does not exacerbate the existing energy deficiency as this has been shown to accelerate bone loss [54]. Thus, the energy cost should be as little as possible because it is unlikely that an athlete will succeed in replacing large amounts of calories when their energy availability is already compromised. If the intervention were to be a burden on time, it is unlikely that athletes would be able to integrate the exercise into their training schedules without compromising the volume of their existing training—which is typically high in endurance sports. The Mechanostat theory states that a loading stimulus must exceed a minimum strain threshold to elicit an adaptive response [55]. Such a threshold is likely higher in WBEA (compared to non-athletes and non-weight bearing athletes) due to years spent performing weight-bearing activity, such that particularly high loading rates are required [55]. Therefore, an exercise intervention needs to be brief, cost as little energy as possible, and elicit a high level of strain, to protect the bones of WBEA suffering from LEA.

A high level of strain can be exerted by high magnitude, low frequency loads applied at a fast rate, such as during jumping and bounding, and can be enhanced further still by incorporating multi-directional impacts [23]. Furthermore, mechanosensitivity to such a stimulus dampens after a relatively small number of loading cycles and is only restored after a re-sensitization period of 8–24 h [23, 56, 57]. In theory, performing low repetition high-impact exercise at low frequency (up to three times per day) represents a model whereby a maximum adaptive response can be achieved in a timely manner. This may depend on an athlete’s ability to plan this exercise and allow time between training sessions to minimize interference between the high-impact stimulus and regular training. It is not yet known how long this would need to be. Athletes would expend a negligible number of calories performing high-impact exercise (making it easy to compensate for), however, even if the energy cost is not fully accounted for, it is possible that the high levels of strain could outweigh the effects of a marginally lower energy availability. Most importantly, the effects of the intervention would need to oppose the effects of LEA on bone, and continue to do so during periods of LEA.

The use of bone (re)modeling markers to monitor the effects of bone treatment has been supported by the International Osteoporosis Foundation as they respond more rapidly than structural measures such as BMD [77]. To investigate a cause and effect relationship between LEA and bone health, research has sought to control energy availability and measure the acute change in systemic markers of bone (re)modeling. In the only study on WBEA, eight male distance runners exercised for three days in energy balance in one condition, and at 50% of energy balance in another. Significant reductions in serum pro-peptide of type 1 collagen (P1NP; a marker of bone formation) were shown in the restricted condition only [78]. Energy availability per se was not quantified and no measures were taken to prevent the reduction in micronutrient availability associated with energy restriction, which will have influenced the data. It has also been shown that P1NP concentrations are significantly reduced in active women following five days at an energy availability of 15kcal·kgLBM⁻¹·day⁻¹; p = 0.01, d = 0.36) when compared to the same duration at 45kcal·kgLBM⁻¹·day⁻¹ [10]. A significant effect was not shown in active men following the same protocol (p = 0.12, d = 0.31), although the effect size was small to moderate in both sexes [10]. In active women, P1NP was reduced after just three days at an energy availability of 15kcal·kgLBM⁻¹·day⁻¹ (p = 0.052, d = 0.36) when compared to the same duration at 45kcal·kgLBM⁻¹·day⁻¹ [79]. These studies by Papageorgiou et al. [10, 79] provided a multivitamin multi-mineral supplement to the restricted groups to maintain micronutrient availability. Findings suggest that both males and females may be affected but females were more sensitive. In support of this, lower levels of P1CP (a formation marker similar to P1NP) were evident in women following five days at a greater energy availability of 30kcal·kgLBM⁻¹·day⁻¹, compared to five days at an energy availability of 45kcal·kgLBM⁻¹·day⁻¹ [80].

Zanker and Swaine [78] showed no significant changes in two urine borne markers of bone resorption in response to 50% energy restriction compared with energy balance in male distance runners. However, energy availability was not reported and differences in micronutrient availability between conditions may have confounded the data. Papageorgiou et al. [10] also showed no effect of five days at an energy availability of 15kcal·kgLBM⁻¹·day⁻¹ on serum β-carboxyl-terminal cross-linked telopeptide of type 1 collagen (β-CTX; a marker of bone resorption) in active men. In active women, β-CTX significantly increased following five days at an energy availability of 15kcal·kgLBM⁻¹·day⁻¹ compared to 45kcal·kgLBM⁻¹·day⁻¹ [10], but not following three days at an energy availability of 15kcal·kgLBM⁻¹·day⁻¹ [79]. Furthermore, a urinary marker of bone resorption significantly decreased following five days only at the most severe level of LEA (10kcal·kgLBM⁻¹·day⁻¹) in women (p < 0.001, d = 0.96); but did not following more moderate levels of energy availability: 20 (p > 0.05, d = 0.16) or 30kcal·kgLBM⁻¹·day⁻¹ (p > 0.05, d = 0.16) [80]. Many different (re)modeling markers exist and heterogeneity between some of the studies makes direct comparisons difficult; however, the findings are consistent with those based on P1NP and β-CTX, which are the recommended international reference markers [77].

These findings show that LEA suppresses markers of bone formation within five days in active females and male WBEA. Females appear to respond with greater sensitivity than males, which suggests that markers of bone formation may also be suppressed in female WBEA in response to LEA, at least to a similar extent as in male WBEA. In male WBEA, it is possible that markers of bone resorption may only increase if LEA persists for longer than five days. Conversely, markers of bone resorption are increased in response to LEA in active females within five days, but only following LEA of greater duration or severity than is required to suppress markers of bone formation. This implies bone formation is more acutely affected by LEA than bone resorption which may lead to net bone breakdown. Due to a lack of evidence, it remains unclear whether the effect on bone resorption is the same in female athletes versus physically active women. Furthermore, in all studies that restrict energy availability, the availability of at least one of the macronutrients (carbohydrate, protein or fat) is inevitably reduced in the restricted condition and it has been shown that β-CTX is increased independent of energy availability in situations of reduced carbohydrate availability [81]. The contribution of reduced macronutrient availability to the effects described could not be determined.

The relationships between acute changes in markers of bone formation and resorption in response to LEA and long-term bone health are unclear. Sustained and localized acceleration of bone remodeling may play a role in the pathogenesis of bone stress injury in athletes [82, 83]. However, bone (re)modeling markers are typically measured systemically and, therefore, any change in the balance of such markers does not necessarily reflect a change in bone remodeling at a specific site. Accordingly, prospective research has shown no significant differences in bone (re)modeling markers between athletes and military recruits who suffered a stress fracture and those who did not [84, 85]. It is commonly reported that a reduction in bone resorption leads to long-term bone accrual; however, in an individual bone remodeling unit, resorptive osteoclast cells initiate the remodeling cycle and precede bone formation [86]. Thus, suppression of resorption might actually inhibit adaptation [87]. Although the acute effects of LEA on bone (re)modeling markers cannot yet be interpreted in terms of long-term bone health, it is reasonable to assume that preventing the acute effects from occurring would be beneficial given that evidence suggests the long-term effect of LEA is detrimental to bone structure, strength and stress injury risk.

That markers of bone (re)modeling respond acutely to LEA offers an exciting avenue to explore whether high-impact interventions modify the bone’s response to LEA, without the methodological challenge of controlling or monitoring diet and exercise for several months (at least). However, the bone (re)modeling response to repeated days of high-impact exercise is poorly understood. The existing data show either no change or a reduction in markers of bone formation and resorption [88, 89], and none of the previous research has simultaneously manipulated energy availability. Also, it is not clear to what degree these findings are due to the interventions per se, or merely a result of biological variation due to poor standardization [86]. One study was conducted in a cohort of hospitalized anorexia nervosa patients, which provides little information regarding whether high-impact exercise might prevent the acute effects of LEA in active populations [89]. Furthermore, the markers measured in response to short-term exercise interventions are often different from those measured in response to short-term LEA, which limits direct comparisons.

Using prolonged moderate impact running exercise, Papageorgiou et al. [79] exposed young active women to three conditions (each 3 days in duration) in a randomized crossover design: LEA (15kcal·kgLBM⁻¹·day⁻¹) with daily exercise, equally LEA non-exercise, and balanced energy availability (45kcal·kgLBM⁻¹·day⁻¹) with no exercise. P1NP was significantly reduced in the LEA non-exercise condition (p < 0.001, d = 0.91) but not in the LEA daily exercise condition (p = 0.14, d = 0.30). There was no effect on β-CTX in any condition [79]. It was suggested that daily moderate impact exercise might have prevented the acute effects of LEA on bone formation in active individuals. However, it is not known whether this effect persists over a longer period or whether such effects translate to WBEA who are more accustomed to the mechanical loading associated with daily prolonged running. As discussed, the effects of LEA on bone (re)modeling markers were more marked after five days than after three, but the independent effect of impact exercise has not been tested beyond a period of three days. Also, the intervention was neither time nor energy efficient (duration and energy expenditure were 129 ± 10 min day⁻¹ and 30kcal·kg⁻¹·day⁻¹) and has not yet been tested in men [79]. Future research could use a similar design to that used by Papageorgiou et al. [79], but preferably use a longer duration of LEA or mimic a typical athlete training week to determine whether low repetition high-impact interventions prevent the acute effects of LEA on markers of bone (re)modeling in male and female WBEA. Findings may be used to improve evidence-informed practice, and inform longitudinal studies designed to investigate the effect of such interventions on long-term bone health and stress injury in athletes at risk of LEA.

WBEA are frequently exposed to episodes of LEA in association with the demands of their sport. In female athletes, including WBEA, LEA is identified as a causative factor in FHA which, in turn, is associated with impaired bone health, including lower BMD, bone mineral content, and trabecular volumetric BMD, thinner cortices, and reduced bone strength. These impairments have been observed at load bearing sites, such as the proximal femur and tibia, and may increase the risk of bone stress injury as well as osteoporotic fracture. Not until recently was the existence of LEA associated bone loss acknowledged in men and, therefore, there are scarce data in male WBEA. High-impact exercise can be highly osteogenic with a low energy cost. Low repetition high-impact interventions have been shown to increase BMD, cortical thickness and estimated strength and preliminary evidence suggests that some of these effects may occur despite LEA. Such interventions may help attenuate bone change in WBEA and reduce the risk of bone stress injury in those at risk of LEA. Research examining the short-term effects of LEA in active men and women suggests that circulating markers of bone formation may be suppressed and circulating markers of bone resorption may be increased within five days, with women possibly responding to LEA with greater sensitivity. Prolonged moderate impact exercise may help mitigate the effects of short-term LEA; however, it is currently unclear whether this would be of benefit to long-term bone health. Given that bone health in WBEA can become compromised due to LEA, investigation of methods which may protect bone health in the face of LEA is of clinical importance.