2.3.1. Thyroid Hormones

The important role of thyroid hormones in determining energy expenditure and basal metabolic rate has long been recognized [49]. With aging, there is a general increase of the incidence of thyroid diseases [50]. Apart from this increase and according to several population studies, aging is associated with subclinical hypothyroidism, i.e., increased levels of thyroid-stimulating hormone (TSH) with free thyroxine (FT4) levels remaining in the normal range [51,52]. Some authors even ascribe a beneficial adaptation of physiological aging to these reduced TSH levels in the elderly by preventing excessive catabolism [53]. While the free triiodothyronine resistance index was negatively associated with aging in males, TSH levels were positively associated with age in females [54]. These results underline a possible sex-specific effect of alterations of thyroid hormones with aging.

Although endocrine effects have to be separated from alterations of body composition and PA behavior with aging, RMR is lower in older individuals, potentially even after adjusting for differences in body composition [43]. The age-related decline of RMR, however, cannot fully be ascribed to alterations of body composition or differences in thyroid hormone status [55].

#### 2.3.2. Hypothalamic Growth Hormone-Insulin-Like Growth Factor-I Axis

This axis includes the secretion of growth hormone (GH; somatotropin) from the somatotropes of the pituitary gland into the circulation, and the successive stimulation of insulin-like growth factor-1 (IGF-1). This endocrine system drives anabolic effects on protein synthesis and growth and hence plays an important role in maintaining muscle mass. In elderly individuals, GH secretion, together with IGF-1 levels decrease starting with the third decade [56]. There is also a reduction in GH releasing hormone (GHRH)- induced GH secretion, likely reflecting changes of neurotransmitter control and reduced

hypothalamic GHRH synthesis related to the aging brain [57]. Regular physical exercise is thought to modulate activity of the GH-IGF-1 axis throughout the lifespan, potentially preserving muscle mass in the elderly. With regard to that, GH responses to a cycling sprint, resistance or endurance exercise bout were compared in young and middle- aged men. While resting GH concentration and objective parameters of exertion were not different between groups, GH response to exercise was greater in the young compared to their older counterparts [58]. When investigating the effect of a 12-week resistance training program on GH and testosterone secretion in young (23 years) and older (63 years) individuals, the authors found that, regardless of age, this training modality elicits GH and testosterone secretion. Response and magnitude, however, was different between the two groups [59]. Mechano growth factor (MGF), a splice variant of the IGF-1 gene, is supposed to be an important local factor promoting satellite cell proliferation in muscle. Although short-term (5-week) resistance exercise failed to increase expression of MGF in elderly as compared to young individuals [60], longer-term training of 12-weeks was still able to upregulate expression of this factor in the elderly [61]. Exercise training can stimulate the GH-IGF-1 axis as well as MGF in the elderly, albeit somewhat less so in older as compared to younger individuals. This may reflect an age-related desensitization to mechanical loading. It has to be noted that short study duration, differences in training motivation or the inability to achieve a sufficient absolute exercise intensity in the elderly as well as a sex-bias towards male study participants may bias these findings.

#### 2.3.3. Hormones Regulating Appetite and Food Intake

Above the age of 65 years, there seems to be a decrease in appetite and food intake, which predisposes to undernutrition. A negative energy balance has implications for chronic disease progression and mortality rate [62]. Hormones mediating the anorexic effect of aging include cholecystokinin (CCK), leptin, and ghrelin. CCK is a gastrointestinal peptide hormone produced by enteroendocrine cells of the duodenum that mediate satiating effects by binding to receptors in the central nervous system. Aging is associated with increased CCK concentrations as well as a greater sensitivity to the satiety-inducing effect of this hormone [63]. In line with that, whey protein ingestion resulted in greater plasma concentrations of CCK and gastric inhibitory peptide in older compared to younger individuals [64]. In addition, the increased activity of the anorexigenic hormone leptin, an adipokine derived from adipose tissue in humans as well as the reduced activity of the orexigenic hormone ghrelin, a gastrointestinal molecule derived from the stomach, seem to further reduce hunger and thereby hamper energy intake in the elderly [65,66]. Obesity is another factor that can dysregulate the endocrine role of leptin secretion from adipocytes, so that hyperleptinemia due to high fat mass fails to negatively regulate food intake, a state termed leptin resistance [67]. A recent study showed that higher fitness levels in older individuals were associated with lower leptin levels and inflammation, regardless of adiposity, suggesting a protective effect of physical fitness towards development of leptin resistance [68]. Nevertheless, the modulatory role of PA on hormones governing appetite and food intake in the elderly remains understudied. In general, individuals with higher levels of PA experience blunted satiety and amplified hunger compared to those being less physically active, likely in order to compensate for the increased PA induced energy expenditure [69]. It seems, however, that chronic exercise can affect perceptions of hunger and energy intake independent of body composition and sex. Individuals being physically active may be more sensitive to regulating energy balance by improved adjustments of energy intake and density of food [70]. The paucity of data on this subject in older athletes of both sexes, however, does not allow definitive conclusions with regard to this age group.

#### 2.3.4. Insulin, Glucose and Metabolic Pathways Mediating Glucose Homeostasis

Insulin is the most important hormone mediating control of glucose homeostasis, i.e., cellular uptake, utilisation and endogenous production. Aging is associated with

deteriorating glucose homeostasis [71]. It has been suggested that, starting from the fourth decade, fasting plasma glucose increases by about 0.055 mmol/L per decade, alongside with a gradual rise of glucose concentrations obtained after 120 min of a 75 g oral glucose tolerance test [72]. Further to that, aging is a risk factor for brain insulin resistance, i.e., impaired sensitivity of central nervous pathways to insulin [73]. Interestingly, high insulin action in the brain influences long-term weight managemen<sup>t</sup> and is associated with a favorable body fat distribution [74]. In general, insulin is secreted in a pulsatile manner. During the basal as well as insulin-stimulated state, both amplitude and number of pulses are reduced with aging [75]. This goes hand in hand with decreased effectiveness in reducing hepatic glucose output and increased liver insulin clearance [76]. Glucagon concentrations, in turn, do not seem to be affected by age [77]. However, the rise in hepatic glucose production after glucagon stimulation, i.e., hepatic sensitivity to glucagon, is increased in elderly individuals compared to their younger counterparts [78].

Seminal studies from the lab of John Holloszy and colleagues point towards the importance of physical training for maintaining glucose tolerance. The researchers studied older lean and overweight untrained as well as trained individuals compared to young untrained and trained individuals. They found that oral glucose tolerance, defined as the area under the glucose curve, was twofold poorer in the untrained older individuals as compared to the other groups. Also, the insulin response to glucose was higher in the untrained groups as compared to the older and younger trained groups [79]. These results are suggestive of reduced glucose tolerance and insulin sensitivity with aging and a counter regulatory role of lifelong exercise training.

A similar, more recent study used the glucose clamp technique to assess insulin sensitivity in younger (age range 24–47 years) and older (age range 60–75 years) athletes as well as younger and older normal-weight and obese individuals. Insulin sensitivity was highest and similar in young and older athletes, followed by normal weight individuals and lowest in obese young and old individuals [80]. This study underlines the fact that body composition and physical inactivity, but not age per se are related to the development of insulin resistance. As glucose disposal decreases over the lifespan, several factors contributing to reduced insulin action have to be considered. These include a rise in total body fat, in particular visceral fat, and reductions of caloric intake, PA and consequent decrements in lean mass or certain drugs and age-related diseases.

In this context, it is important to also scrutinize cellular mechanisms responsible for age-induced reductions in insulin-stimulated glucose uptake. In brief, upon binding to its receptor, insulin stimulates autophosphorylation of the insulin receptor, tyrosine phosphorylation of the insulin receptor substrate-1 (IRS-1) and association with phosphoinositide 3-kinase (PI3K) and subsequent phosphorylation of Akt2 at serine 473 and threonine 308 sites and AS160 phosphorylation, which finally promotes glucose transporter type 4 (GLUT4) translocation to the plasma membrane to stimulate glucose uptake [81]. Aging can selectively affect elements of this signalling cascade. For instance, Akt phosphorylation by insulin was ~40% lower in healthy, lean, elderly individuals compared to young body mass index (BMI)-, fat mass- and habitual PA-matched volunteers [82]. This was in line with a 25% reduction in insulin-stimulated rates of muscle glucose uptake. Another study found that AS160 phosphorylation under insulin-stimulated conditions was reduced, together with ~30% lower whole-body insulin sensitivity, in aged compared to young individuals [83]. Participation in an exercise training program of either strength or endurance exercise improved whole-body insulin sensitivity and abrogated impairments in AS160 phosphorylation in these individuals. Different factors have been discussed that could contribute to this reduction in insulin action in elderly individuals. Among these, a reduction in oxidative capacity [84], impaired mitochondrial metabolic flexibility during insulin stimulation [82], oxidative stress [85] or dysfunctional regulation of mitochondrial dynamics [86] may contribute to alterations of mitochondria with aging. These changes in turn promote increased deposition of ectopic lipids within the myocyte as intramyocellular lipids (IMCL). This site of lipid deposition has repeatedly been shown to be particularly

detrimental to insulin action [87]. Compared to younger volunteers, IMCL content, assessed by 1H magnetic resonance spectroscopy, was 70% higher in elderly individuals [82]. Decrements in mitochondrial activity together with excess caloric intake promote lipid synthesis and storage and yield bioactive diacylglycerols, which are known to accumulate within the plasma membrane, where they recruit novel protein kinase C isoforms that facilitate inhibitory phosphorylation of IRS1 [81]. Ceramides, another important class of bioactive lipids, may inhibit Akt phosphorylation [81]. These intracellular processes impair the insulin signalling pathway and eventually contribute to impaired insulin-stimulated cellular glucose uptake.

In summary, aging is associated with deteriorating glucose homeostasis. While insulin secretion is diminished over the lifespan, glucagon action is only marginally affected. Besides decrements in hormones mediating control of glucose homeostasis, decreased PA, together with reduced mitochondrial function will promote ectopic lipid deposition in muscle, which further impairs insulin signalling and glucose uptake.

## *2.4. Bone and Bone Metabolism*

Bone is a hard tissue that derives its tensile stiffness from the organic extracellular matrix and its stiffness in shear and compression from an inorganic calcium-phosphate compound (so-called bone apatite). Whilst the majority of the organic matrix consists of type 1 collagen (>90%), there are also other collagens as well as other protein constituents, such as proteoglycans, carboxylated osteocalcin, osteopontin and TGF-β (Transforming growth factor beta (TGF-β)). The role of these latter constituents is only just evolving and seems to be primarily related to endocrine and regulatory functions, rather than serving mechanical purposes. Four different types of cells govern biological activity in bone. Ostoeblasts lay down bone protein onto existing bone surfaces, and they also initiate mineralization. When such bone formation drifts cease, osteocytes either become dormant as bone lining cells, or they transform into osteocytes that are fully covered by bone matrix and thus remote to the bone's surface. However, osteocytes still communicate via gap junctions with other osteocytes and lining cells, which allows trafficking of electrolytes and small organic compounds. Osteocytes are also thought to sense mechanical strains, to transfer that information into biological signals that help to adapt bone to its mechanical purposes, and also to initiate the repair of bone microdamage through the process of bone remodeling [88,89]. Osteoclasts, finally, trans-differentiate from peripheral blood mononuclear cells and have capacity to dissolve the inorganic matrix through acidification, and subsequently also the protein phase in order to resorb bone tissue.

The four types of bone cells work in co-operation with each other. Thus, the activity of osteoclasts is under the control of osteoblasts through receptor activator of NFKB ligand (RANKL), as well as under the control of osteocytes by sclerostin [90]. In addition, TGF-β from the bone matrix stimulates activation of osteoblasts [91]. The reciprocal control of osteoblastic activity by osteoclasts is less well established, but may involve compounds such as ephrinB2 and other factors [92]. These cell-cell interactions seem fundamental for the bone remodeling process [93]. Herein, osteoclastic bone resorption precedes osteoblastic bone formation, a process that is often triggered by material microdamage [94], thus helping to repair damaged bone. Modeling, on the other hand, serves the purpose of 'shaping' bones through formation and resorption that occurs on opposing surfaces of the bone, thus resulting in bone 'drifts' [95]. Thus, remodeling and modeling are two diverse processes. Whilst modeling is the main contributor to bone turn-over at young age, the bone turn-over after closure of the growth plate is mostly due to remodeling. With regards to aging, it is also noteworthy that bone turn-over is typically greater in young than in older people [96]. This is partly due to a slowing of the remodeling process, i.e., the entire cycle from activation of osteoclasts until closure of the resorption cavity, which takes approximately 90 days in young people but 120 days in old people [93]. In addition, it has been demonstrated that the ability to respond to bone microdamage becomes blunted with increasing age [97].

It is important to understand that the largest forces that bones experience emerge from regional muscle forces and impact loading, rather than from body weight per se. As a result, muscle mass and bone mass are well-adapted to each other [98]. This mutual relationship between muscle and bone is largely driven through the bone's response to local tissue strains [99]. Thus, the mechanostat theory proposes that bone counteracts excessive deformations with enhanced bone formation, whereas sub-threshold deformations are answered by streamlining the bone's structure [100]. In addition, evidence suggests also an independent effect of strain rate on bone accrual. There are also several signaling pathways that interact with muscle and bone. For example, bone morphogenetic protein 1 (BMP1) has been shown to positively affect muscle growth in mice [101]. TGF-β, which initially had been known for its stimulating role for osteoblasts, has now been recognized to negatively affect calcium handling within skeletal muscle, and to thereby hamper muscles electro-mechanical coupling [102]. Moreover, undercarboxylated osteocalcin from the bone enhances insulin production, cognitive functioning and muscular fuel utilization [103]. Thus, there are several compounds from bone's non-collagen protein phase that unfold important systemic actions on skeletal muscle and athletic performance.

In addition to the mechanical determinants, bone is also under control of several endocrine systems. Both sex hormones have important effects, with the estrogens and androgens being more effective on the endocortical and periosteal surfaces, respectively [104]. From an evolutionary perspective, estrogens may serve the role of 'packing' bone minerals that are not mechanically required [98], so that sufficient calcium depots are available when breastfeeding taps on the calcium resources [105]. In this regard, the complete withdrawal of estrogen after menopause is a catastrophic event for the female skeleton, leading to bone loss and an incidence rate for vertebral and femoral fractures of 1% per year, respectively [106]. In female endurance athletes, the so-called female athlete triad may occur, where underfeeding is associated with amenorrhea, low bone mineral density and fractures, even at young age [107]. In addition to sex steroids, there are also other hormones impinging on bone, such as parathyroid hormone, the D-hormone and insulin-like growth factors. Bones are responsive to exercise [108], which is primarily thought to be through mechanical effects. Rapid running and jumping elicits larger strains and strain rates than static exercises [109], and power athletes have stronger leg bones than endurance athletes [110–112]. The largest effects have been observed for the humerus, the structure of which can be twice as strong in tennis and baseball players in the active as compared in the passive arm [113,114]. Sideto-side differences in the legs of jumpers are much smaller than in the arm [115], which could be due to the fact that jumpers load both legs substantially during run-ups, whereas tennis players place only small loads on their passive arms.

With advancing age, bone mass is lost in the general population, a phenomenon that is much more pronounced in the spine and in the upper extremity than in the legs [116]. Bone benefits acquired through exercise at young age can still persist several decades after cessation of exercise training [117]. Conversely, when taking up an exercise activity during adulthood, i.e., after fusion of the growth plates, then it is still possible to moderately enhance bone mass and bone strength in the loaded bones, as demonstrated by observations in masters tennis players achieved in adulthood [118]. As to another question, namely what happens to masters athletes if exercise is continued across the life span, results from cross-sectional studies sugges<sup>t</sup> that the bone benefits attained at young age [119] persist into old age, although they may fade away with advancing age [120]. A recent longitudinal study in 71 track and field masters with 4-year follow-up, however, demonstrates that tibia bone mass and strength can be preserved, or even increased in male power athletes, whilst male endurance athletes and female track-and-field masters lose tibia epiphyseal bone mass at a rate of approximately 5% per decade [38]. It has to be considered, however, that some of the female masters athletes in that study were aged between 50 and 60 years, i.e., within the perimenopausal decade, and that this study could not discern between the effects of aging and of menopause. It therefore seems fair to conclude that engagemen<sup>t</sup>

in track and field events such as sprint running and jumping has potential to maintain or increase bone health into old age.

#### **3. Performance and Trainability of Older Athletes**

*3.1. Performance Decline in the Aging Athlete*

Declines in cardiovascular fitness and physical performance are the main features of aging human beings [121,122]. The decline starts in the third to fourth decade and depends on genetic and life-style characteristics, in particular on PA through life-span. High cardiorespiratory fitness (CRF: aerobic capacity, VO2max) and regular PA are associated with longevity even when adjusted for relevant confounders [122,123]. Differences of the slope in performance decline between aging athletes and their sedentary peers are still a matter of debate, but absolute performance levels remain considerably high in masters athletes of both sexes at all ages [28]. Whereas most studies primarily focused on agedependent changes in running and swimming performance [124,125], only a few have considered a broader range of sports [126]. Data derived from mean winning performance times in track and field events of 10,000 participants in Senior Olympic Games demonstrate a slow decline from the age of 50 to 75 years which became dramatically steeper after the age of 75 [126]. Those findings have been expanded by a recent study suggesting an additional steepening of the endurance performance (in track and field events) in very old masters athletes aged 80+ [127]. Whereas no differences in performance decline in sprint versus endurance events were found for men, sprint performance declined more steeply than that of endurance in women [128]. However, when accounting for kinetic energy contributions to the metabolic costs, sprint and distance running world records depicted remarkably similar age trends [129]. A similar decline was shown in mountain runners from 50 to 70+ years; this decline was, however, slightly more pronounced in females compared to males [28]. Although the performance decline in this study seemed to be slightly steeper in masters athletes than in sedentary persons of the general population, it is important to note that endurance performance remained about 3.5 times higher in older athletes [28]. A recent review confirmed, based on current data from masters athletes, that most of the physiological mechanisms determining VO2max (i.e., pulmonary and cardiovascular function, blood oxygen transport capacity, skeletal muscle capillary density and oxidative capacity) are profoundly modulated by regular PA during the entire life-span [9]. In contrast to cardiovascular and skeletal muscle adaptations to PA and exercise training, pulmonary adaptations seem to play a rather minor role for the maintenance of performance in the aging athletes [9,10]. In summary, all these findings indicate that endurance performance (i.e., VO2max) inevitably declines when getting older but can be favorably modified by regular PA and especially by exercise training at all ages in both sexes (Figure 3).

## *3.2. Sex Differences in Performance, Decline Rates, and Training Effects* 3.2.1.AerobicFitness

Upper limits of VO2max values can exceed 90 mL/min/kg in young male elite athletes [134] compared to about 35 to 37 mL/min/kg in the general population aged 35–45 years [135]. Above the third age decade, VO2max values decline by approximately 10% per decade but this decline may be considerably modulated by training [9,10]. Extraordinarily high aerobic capacity has been reported in lifelong physically active very old individuals of both sexes. For instance, VO2max values of 38 (±1) vs. 21 (±1) mL/min/kg were shown for 81-year-old male endurance athletes (*n* = 9) vs. age-matched healthy untrained persons [136]. Even more impressive, a VO2max of 42.3 mL/min/kg has been recently reported in a 83-year old female masters runner [137]. On top, a 101-yr old athlete improved his VO2max following 2 years of specific training (to 103 years of age) from 31 to 35 mL/min/kg [138]. Baker and Tang compared the aging-related rates in performance decline of masters records of both sexes [126]. The age of males/females when performance had declined to 50% of the maximum performance at 30–35 years were 90/84 years for walking, 87/84 years for swimming, 78/72 years for jumping, and 74/60 years for

weight lifting [126]. The larger sex differences in weight lifting compared to walking or swimming might be attributable to differences between lower and upper body performance (composition). While 12% sex difference was reported for running, swimming and cycling [139,140], the sex difference was 17% in cross country skiing (skating technique), again indicating different lower and upper body performance between sexes [135]. The relative (to body mass) VO2max differs by about 10% between male and female athletes, which is primarily explained by the higher percentage of body fat and lower hemoglobin levels in women [135,141]. Although this indicates similar overall effectiveness of the oxygen delivery and utilization systems in men and women, specific differences in cardiac and skeletal muscle adaptations exist. With regard to cardiac adaptation to regular exercise training, greater LV end-diastolic cavity sizes and LV mass are observed in trained versus untrained individuals, which is more pronounced in male athletes [142]. Sex differences in LV mass cannot completely be explained by the lower body size of the female athletes, but may at least partly result in the larger blood pressure increase at peak exercise in men [143]. Performance in sprint cycling (200 m) was effectively halved (from the maximum performance at 30–35 years) at an age of 80 years in males compared to an age of 59 years in female masters athletes [126]. Regular exercise training of masters athletes can prevent the shrinking of type I muscle fibers [144], and minimize loss of type II fibers, and it may also prevent fiber type grouping [145] following denervation-reinnervation cycles with aging in both sexes. Notably, females seem to develop a slower muscle fiber phenotype due to progressive slowing of discharge rates [146], likely explaining the relatively large sex difference in the decline of sprint cycling performance [126] (see below).

**Figure 3.** Schematic presentation of the age-related performance decline, i.e., in aerobic capacity (VO2max) and maximal grip strength, for both sexes, depending on physical training status, ranging from well-trained to sedentary. The figure is based on the meta-analysis from Fitzgerald et al. (1997) [130], with contributions from Burtscher et al. (2008) [28], Booth et al. (2010) [131], Landi et al. (2017) [132], and (defining frailty threshold as VO2max of 18 mL/min/kg) Carr et al. (2006) [133].
