Role of the nervous system in sarcopenia and muscle atrophy with aging: strength training as a countermeasure


Corresponding author: Professor Per Aagaard, PhD, Institute of Sports Science and Clinical Biomechanics, University of Southern Denmark, Campusvej 55, DK-5230, Odense, Denmark. Tel: +45 6550 3876, Fax: +45 6550 3480, E-mail:


Aging is characterized by loss of spinal motor neurons (MNs) due to apoptosis, reduced insulin-like growth factor I signaling, elevated amounts of circulating cytokines, and increased cell oxidative stress. The age-related loss of spinal MNs is paralleled by a reduction in muscle fiber number and size (sarcopenia), resulting in impaired mechanical muscle performance that in turn leads to a reduced functional capacity during everyday tasks. Concurrently, maximum muscle strength, power, and rate of force development are decreased with aging, even in highly trained master athletes. The impairment in muscle mechanical function is accompanied and partly caused by an age-related loss in neuromuscular function that comprise changes in maximal MN firing frequency, agonist muscle activation, antagonist muscle coactivation, force steadiness, and spinal inhibitory circuitry. Strength training appears to elicit effective countermeasures in elderly individuals even at a very old age (>80 years) by evoking muscle hypertrophy along with substantial changes in neuromuscular function, respectively. Notably, the training-induced changes in muscle mass and nervous system function leads to an improved functional capacity during activities of daily living.

This article provides a review of the alteration in neuromuscular organization and function induced by aging, while addressing its relationship to muscle atrophy and sarcopenia at old age. In addition, the article discusses the use of strength training as a countermeasure for regaining neuromuscular capacity and function in the aging individual. For the interested reader, several informative review articles have been published over the last decade that describe the remodeling in neural organization and its coupling to muscle tissue maintenance with aging in both human (Grounds 2002; Vandervoort 2002; Enoka et al., 2003; Grounds et al., 2008) and experimental animal settings (Luff 1998; Tam & Gordon 2003a; Edström et al., 2007), respectively.

Neuromuscular changes with aging

Aging is characterized by a gradual loss of spinal motor neurons (MNs) due to apoptosis, reduced insulin-like growth factor I (IGF-1) signaling, elevated amounts of circulating cytokines [tumor necrosis factor (TNF)-α, TNF-β, interleukin (IL)-6] as well as increased cell oxidative stress. Some, but not all, denervated muscle fibers are reinnervated through collateral sprouting of nearby surviving motor axons or motor end plates, which results in the formation of very large motor units (MUs) (discussed in detail below). Consequently, force steadiness and fine motor control are impaired with aging (Tracy & Enoka 2002), although this change may relate most directly to age-associated changes in the variability and common modulation of MU discharge rate in both agonist and antagonist muscles (for a detailed review on neural aspects related to force steadiness and aging, see Enoka et al., 2003).

The age-related loss of spinal MNs leads to a decline in muscle fiber number and size (sarcopenia), resulting in impaired mechanical muscle performance (reduced maximal muscle strength, power, and rate of force development [RFD]) that translates into a reduced functional capacity during everyday tasks (level walking, stair walking, rising from a chair). Specifically, muscle atrophy may be considered the endpoint of different conditions such as sarcopenia, cachexia, and pre-cachexia status. All three conditions are characterized by muscle atrophy (as well as immobilization) but the underlying mechanisms are likely, different although, aging is often common to all of them. Aging is accompanied by a substantial reorganization in the neuromuscular system and the CNS, which contributes to the loss in motor performance.

As discussed in more detail below, age-related alterations in neural function can be identified both at the peripheral level (axons, motor end plates) as well as at spinal and supraspinal levels. Specifically, experimental evidence of neuromuscular remodeling at old age include reports of very large MUs based on macro-electromyography (EMG) recording (Stålberg & Fawcett 1982; De Koning et al., 1988; Stålberg et al., 1989; Masakado et al., 1994) and findings of fiber-type grouping obtained by muscle biopsy sampling (Lexell & Downham 1991; Andersen 2003). Further, old subjects show compressed rate coding during graded muscle contraction compared with young individuals (Barry et al., 2007), while resting H-reflex excitability (Hmax/Mmax) appears to be reduced in elderly individuals (Scaglioni et al., 2002) indicating elevated presynaptic inhibition of Ia afferents (Morita et al., 1995) and/or reduced MN excitability in the resting condition. Reduced levels of central activation (CA) (for a review, see Klass et al., 2007) and elevated levels of antagonist muscle coactivation during MVC (Häkkinen et al., 1998a, 2000; Izquierdo et al., 1999; Macaluso et al., 2002) and in postural stability tasks (Earles et al., 2000) can be observed in some (but not all) elderly individuals. In addition, aging is accompanied by cortical reorganization (Ward & Frackowiak 2003; Ward et al., 2008) and findings of reduced TMS-evoked motor potentials suggest that the excitability of efferent corticospinal pathways is reduced at old age (Eisen et al., 1991; Sale & Semmler 2005).

Spinal MNs and MUs

Electron micrographs obtained in peripheral nerves of aging experimental animals (old rats) show connective tissue infiltration along with decreased myelination and reduced axon density (for a review, see Edström et al., 2007). In human autopsy studies, aging has been accompanied by a reduced number and diameter of myelinated MN axons in the ventral roots, with an accelerated loss of large-diameter axons (Kawamura et al., 1977a, b; Mittal & Logmani 1987) (Fig. 1). In addition, axonal conduction speed is reduced with aging, likely due to drop outs of the largest axonal fibers, reduced myelination, and reduced internodal length (Doherty et al., 1993; Metter et al., 1998; Scaglioni et al., 2002). Findings of increased H-reflex latency in elderly subjects (Scaglioni et al., 2002; Kido et al., 2004) further supports that peripheral nerve conduction speed is reduced at old age. The finding that latency of the direct M wave (5.18 vs 5.19 ms) did not differ between old (68 years) and young (29 years) subjects despite an elongated soleus (SOL) H-reflex latency (33.24 vs 29.85 ms) suggests that aging may affect sensory afferent axons and/or synapses more severely than efferent motor axons (Scaglioni et al., 2002). However, both sensory axons and motor axons may demonstrate degenerative changes with aging (Adinolfi et al., 1991; Lascelles & Thomas 1966).

Figure 1.

 Total number of axonal fibers (top right panel) and number of large diameter axons (diameter ≥12.5 μm, bottom right panel) in right cervical ventral nerve root fibers (C8) of young and old humans (n=10, 21–94 years). Inserted pictures show nerve electron micrographs obtained in young (top left) and old subjects (bottom left) (from Mittal & Logmani 1987).

At the MU level, animal data have demonstrated a 30–40% decline in the number of MUs when adult and old animals were compared (Edström & Larsson 1987; Einsiedel & Luff 1992). Likewise, a reduced number of spinal MNs have been observed in aging humans, where a 25% average loss in the number of spinal MNs were seen at lumbospinal segments L1-S3 over the age range from 20 to 90 years, with several subjects of 60+ years age demonstrating 50% fewer MNs compared with 20–40-year-old subjects (Tomlinson & Irving 1977).

Analysis of the composite surface EMG signal during isometric force ramps of varying contraction intensity using spike-triggered averaging techniques or incremental peripheral nerve stimulation have been used in humans to estimate the number of functioning MUs across the age span (McNeil et al., 2005; Dalton et al., 2008). It should be recognized, however, that the electrophysiological methods used to estimate the number of functioning MUs produce results that are highly dependent on the specific experimental conditions. Thus, motor unit number estimates (MUNE) varies substantially when assessed at low vs high contraction intensities, because high-intensity contractions may preferentially sample large MUs resulting in a smaller MUNE, whereas low-intensity contraction will mainly involve smaller MUs to yield a larger MUNE (Dalton et al., 2008). Consequently, the use of MUNE techniques to compare MU number in young and old individuals is only valid if the methodological limitations apply equally to the two age groups, which may not always be the case.

Early studies have suggested that a marked reduction in the number of excitable (i.e. functioning) MUs may occur from 60 to 70 years of age (Campbell et al., 1973; Brown et al., 1988; Doherty et al., 1993). For instance, in 60–96-year-old individuals the estimated number of MUs in selected lower extremity muscles (extensor digitorum brevis) corresponded to about 30% of the number observed in younger individuals (3–60 years) (Campbell et al., 1973) (Fig. 3). Similar observations have been reported in animal models, where the number of spinal MNs in lower limb muscles decreased at an accelerated rate during the final 1/3 of the life span (Kanda et al., 1996) [cf. Fig. (3b)]. In humans, a reduced number of MUs were observed also for selected upper extremity muscles at increasing age (Brown 1972; Sica et al., 1974; Brown et al., 1988; Doherty & Brown 1993; Galea 1996). More recently, Rice and colleagues observed a 39% decrease in the number of MUs in the TA muscle of old (66 years, range 61–69) compared with young individuals (27 years, range 23–32) along with an even greater decline (61%) in very old subjects (82 years, range 80–89) (McNeil et al., 2005) (Fig. 2, top panel). Notably, maximal isometric muscle strength did not differ between young and old individuals but was reduced in very old subjects (Fig. 2, bottom panel) possibly due to the maintenance of muscle fibers (muscle mass) through the process of collateral reinnervation in the early stages of MU loss (McNeil et al., 2005). It was suggested that the progressive loss of MUs with aging might not lead to functional impairments until a certain critical threshold is reached. For the TA muscle this critical threshold appears to be encountered between 70 and 80 years of age where the number of MUs is about 50% of that found in young, healthy individuals (Fig. 2) (McNeil et al., 2005). The loss in MU number with aging probably is both muscle specific and age-threshold dependent. Thus, estimated MU number in the SOL muscle did not differ statistically between old (∼75 years) and young (∼27years) males (Dalton et al., 2008). In contrast, the estimated number of SOL MUs was higher in young-to-middle-aged persons (5–50 years) compared with very old individuals (90 years) (Vandervoort & McComas 1986). These data suggest that in aging humans the accelerated decline in MU number may occur at a slightly older age for the SOL than for the TA muscle (Dalton et al., 2008). It is possible that this disparity is related to muscle-specific differences in muscle fiber-type composition and/or reflects long-term adaptation to a preferential muscle activity pattern during gait at old age (e.g. greater relative reliance on ankle plantar flexor function), although possibly also arising as a result of the methodological constraints discussed above.

Figure 3.

 (a) Number of spinal medial gastrocnemius (MG) and ulnar motorneurones (MNs) measured in middle-aged and old rats. Old rats demonstrated fewer MG MNs whereas the number of ulnar MNs did not differ between middle-aged and old rats (data from Hashizume & Kanda 1995). (b) Number of MG MNs innervating the lower leg in rats of varying age. The number of spinal MG MNs was significantly reduced in animals aged 28 and 30 months compared with 17-month-old animals (data from Kanda et al., 1996). Open circles refer to data obtained in previous experiments from the same research group (see Kanda et al., 1996). Error bars: group mean SD.

Figure 2.

 Estimated number of functioning motor units (MUs) in the anterior tibial (TA) muscle (top panel) and maximal isometric TA muscle strength (MVC; bottom panel) obtained in young (27 years, range 23–32), old (66 years, range 61–69) and very old subjects (82 years, range 80–89) (data from McNeil et al., 2005).

Influence of IGF-1 on MNs and axons

The endocrine and paracrine production of IGF-1 is reduced in elderly individuals, which may bear significant implications for the prevention as well as the adaptive compensation for the loss of spinal MNs with aging. IGF-1 has potent effects on motor axon myelination, MN apoptosis, stimulation of axonal sprouting, and repair of damaged axons (for a review, see Grounds 2002). However, it remains to be determined to which extent local (paracrine) muscle IGF-1 isoforms produced by myofibers and/or ECM fibroblasts directly helps to maintain spinal MN integrity and function in aging human individuals (Grounds 2002). Elevated levels of inflammatory cytokines TNF-α and TNF-β, which is typically observed in old individuals, can blunt the IGF-1-mediated effects in muscle tissue (Grounds et al., 2008) and might have similar effects in MNs and axons. In addition, cytokines such as IL-6 and IL-1 may further accelerate the loss in muscle mass (sarcopenia) at advancing age.

Animal experiments studying MN disease have shown that IGF-1 facilitates collateral sprouting of surviving axons to innervate denervated muscle fibers, and reinnervation of muscle fibers via axonal sprouting can compensate for the loss of MNs until approximately 50% of the MNs have died (Hantai et al., 1995). Interestingly, these animal data may explain why maximal muscle strength (MVC) was similar in young and old humans despite a ∼40% reduction in MU number whereas a ∼60% reduction in MU number in very old subjects was accompanied by substantial impairments in MVC (Fig. 2). This suggests that a critical threshold of 50% loss in MU number may exists for a maintained capacity for reinnervation of abandoned muscle fibers by means of axonal sprouting (McNeil et al., 2005). Consequently, when axonal sprouting becomes inadequate or absent at old age (discussed below), the gradual loss of MUs results in a substantial loss of muscle bers that produce marked decreases in muscle mass and strength, eventually leading to functional disability (McNeil et al., 2005).

Loss of synaptic MN input, axonal degeneration

In rodents the age-related loss in MNs seems to be small (10–15%) even in advanced age. However, aging MNs show changes (increases or reductions) in dendritic tree size and decreased synaptic input, which appears to divide the MN population in one part that is extensively affected and a subpopulation of more preserved MNs (for a review of the animal model, see Edström et al., 2007). As suggested by Edström et al. (2007) the loss of synaptic input in terms of stripping of afferent boutons from spinal MNs seems to be highly selective and is associated with signs of neuroaxonal degeneration. An effect of reactive oxygen species have been implicated as axons and axon terminals with signs of neurodegeneration and/or neuroaxonal dystrophy have been observed to contain increased levels of total glutathione indicating redox stress (Ramirez-Leon et al., 1999). Further, it has been suggested that axon impairment during aging may start as a distal process (cf. neuroaxonal dystrophy), causing a partial denervation of myofibers within a MU while intact axon branches strive to reinnervate surrounding denervated fibers (Edström et al., 2007). At old age the process of axonal reinnervation appears to be impaired, which is likely to contribute to the reduced regenerative capacity of myofibers observed in sarcopenic muscle (discussed below).

Animal studies have shown that age-related axon lesions are more frequent in ventral roots and peripheral nerves of the lumbar spinal cord compared with the cervical region (Johnson et al., 1995), suggesting that lower limb muscles may be more severely affected by age-related deinervation than upper limb muscles. In support of this notion, a reduced number of MNs was observed in aged rats for the hind limb medial gastrocnemius (MG) muscle, whereas the number of MNs innervating the forelimbs did not differ between middle-aged and old rats (Hashizume & Kanda 1995) [Fig. 3(a)]. These findings are in accordance with the observation in humans that aging is accompanied by a reduced number of muscle fibers in the lower limb (m. quadriceps femoris; Lexell et al., 1988) [Fig. 4b)] while the estimated number of myofibers may remain largely unaffected by aging in selected arm muscles (m. biceps brachii) (Klein et al., 2003).

Figure 4.

 The effect of aging on anatomical muscle size and muscle fiber number. Aging is characterized by a decrease in anatomical muscle cross-sectional area (CSA) and infiltration of non-contractile tissue such as collagen and fat (a) (Caserotti et al., 2008a, b). In addition, the number of muscle fibers is reduced with aging as indicated by autopsy studies (b) (m. quadriceps femoris; data from Lexell et al., 1988).

Data obtained in animal models suggest that although skeletal muscles show loss of fibers and fiber atrophy in advanced age (discussed in detail below), concurrent signs of muscle regeneration can be observed where some markers of regeneration are found to change in parallel with the loss in muscle mass. Notably, sarcopenic muscle may not per se be in a state of poor adaptive responsiveness, rather an impaired capacity for axonal reinnervation of deinnervated myofibers may contribute to the net loss of muscle mass with advancing age (discussed below).

Reinnervation of muscle fibers by axonal sprouting

Perisynaptic Schwann cells at the neuromuscular junction extend processes that bridge between denervated and reinnervated endplates, and guide axonal sprouts to reinnervate the denervated endplates (for a review, see Tam & Gordon 2003a). Observations in aged animal muscle suggest that the capacity for axonal and endplate sprouting is relatively limited (see Luff 1998). Thus, motor endplate sprouting was reduced in the EDL muscle of aged rats in response to partial denervation, suggesting that in highly aged animals nerve terminals and axons may approach their maximal capacity for sprouting (Rosenheimer 1990). Interestingly, in extensively denervated rat muscle the production of perisynaptic Schwann cell processes was inhibited and axonal sprouting was impaired or fully abolished in situations of high neuromuscular activity such as extensive running exercise (Tam & Gordon 2003b). In a rat model of partial TA denervation, high levels of daily neuromuscular activity (running exercise 8 h daily) was shown to inhibit the outgrowth of sprouts by preventing Schwann cell bridging (Tam & Gordon 2003b). It was therefore suggested that with very high daily neuromuscular activity, the large levels of acetylcholine released from intact endplate terminals could suppress glial fibrillary acidic protein synthesis in perisynaptic Schwann cells leading to impaired axonal growth between innervated and denervated terminals (Tam & Gordon 2003b). Similarly to aging, neuromuscular inactivity also seems to produce inhibition of axonal sprouting in partially denervated muscle, likely due to failure of perisynaptic Schwann cell processes to bridge and guide axonal sprouts to reinnervate denervated endplates, and/or caused by a reduced calcium influx into nerve terminals to support sprout growth (see Tam & Gordon 2003a). Thus, it is possible that both chronic inactivity and neuromuscular hyperactivity (i.e. excessive endurance training) may decrease muscle fiber reinnervation in elderly individuals. These findings coupled with the acute rise in myofibrillar protein synthesis rate with electrical stimulation mimicking strength training but not endurance training suggest that resistance rather than endurance exercise (Atherton et al., 2005) may be the preferred training modality for sarcopenic elderly individuals.

In summary, normal aging appears to be characterized by a temporal progression of MU loss accompanied by adaptive axonal sprouting, which is gradually replaced by maladaptive peripheral sprouting. In turn, the gradual impairment in axonal sprouting capacity at old age results in a progressive loss of motor endplate terminals, which initiates an accelerated loss of muscle fibers.

Spinal circuitry function

In humans the effect of aging on spinal circuitry function has been examined by recording of evoked spinal responses (H-reflex) at rest and in various contraction tasks. Old subjects demonstrate reduced H-reflex responses compared with younger individuals when recorded at rest (Morita et al., 1995; Scaglioni et al., 2002), during quiet standing (Koceja et al., 1995; Earles et al., 2000; Koceja & Mynark 2000; Kido et al., 2004), in walking (Kido et al., 2004) as well as during more complex postural tasks (Earles et al., 2000), collectively suggesting that aging may be accompanied by reduced spinal MN excitability and/or increased presynaptic and/or postsynaptic spinal inhibition (Fig. 5). Alternatively, the observed changes in H-reflex amplitude with aging could arise from systematic changes in the MH recruitment curve hence potentially reflecting changes in the recruitment of Ia afferents with this experimental technique. Therefore, findings of age-related alterations in the ability to modulate the H-reflex response during specific motor tasks (isometric contraction, walking) relative to a reference condition (i.e. sitting rest) may provide a more reliable picture of the change in spinal circuitry function with aging. Using this analytical approach, reduced spinal reciprocal inhibition was observed in old individuals during quiet standing and walking (Kido et al., 2004) (Fig. 6), which indicates that human aging involves a remodeling in spinal circuitry function, at least for MNs innervating muscles of the lower limb.

Figure 5.

 Changes in H-reflex excitability with aging. (a) Soleus Hmax and Mmax amplitudes measured during quiet standing with a soleus electromyography level of 15% MVC. Note the decline in Hmax and Mmax with increasing age (Mmaxr=−0.62, Hmaxr=−0.76; P<0.05). (b) Hmax/Mmax ratio showing a significant decrease with aging (r=−0.79; P<0.01) (graphs from Kido et al., 2004).

Figure 6.

 Reciprocal inhibition of ongoing electromyography (EMG) activity elicited by antagonist muscle nerve stimulation (1.5 times motor threshold). (a) Reciprocal inhibition of soleus EMG activity during standing and walking. Inhibition calculated as the difference between the mean rectified EMG of a selected 10-ms window and background EMG, normalized by the background EMG and averaged for a background EMG level of 20–60% MVC. Reciprocal inhibition of the SOL muscle declined with aging (standing: r=−0.57, walking: r=−0.61; P<0.05). (b) Reciprocal inhibition of TA EMG activity, normalized and averaged for the background EMG level of 10–50% MVC. Reciprocal inhibition of the TA muscle declined with aging (standing: r=−0.47, walking: r=−0.75; P<0.05) (graphs from Kido et al., 2004).

Effects of physical activity on the age-related loss of MNs

A relevant question is whether the age-related reduction in the number of spinal MNs and peripheral motor axons is preventable by training or regular physical exercise? As discussed below, the answer seems to be a strict “no.” On the other hand, certain types of exercise (i.e. strength training) can evoke adaptive changes in muscle and CNS function that to a large extent can compensate for the age-induced loss of MNs (discussed below).

In animal models increased muscle loading activity have not prevented the loss in spinal MN number with aging, at least when increased muscle loading was achieved by synergist muscle ablation (Kanda et al., 1996) [Fig. 3(b)]. Likewise, soma size of both α-MNs and γ-MNs remained unaffected by increased muscle loading, suggesting that the beneficial effects of elevated loading activity might be canceled by concurrent detrimental effects (Kanda et al., 1996). Data exist to suggest that neuronal survival may, at least partly, depend on the supply of trophic substance from the CNS (i.e. nerve growth factor) and/or target organ (Kromer 1987). Uptake and transport of trophic substance might depend on neuronal activity (Heuser & Reese 1973), and it is possible that increased muscle activity augments the supply of trophic substances that are transported retrogradely either from distal axon sites or the target motor endplate and/or orthogradely through presynaptic neurons (Kanda et al., 1996). Thus, elevated neuromuscular activity (i.e. exercise) may lengthen the life span of neurons, while in contrast the increase in cellular metabolism may increase the formation of free radicals, which may damage various cell functions and eventually cause cell death (Kanda et al., 1996). This tentative hypothesis is supported by the finding that reinnervation of denervated muscle fibers by means of collateral axonal sprouting was impaired not only in conditions of chronic inactivity but also with neuromuscular hyperactivity as discussed above.

Changes in muscle mass with aging (sarcopenia)

Aging is associated with a loss in muscle mass that results from a reduced number of muscle fibers [Fig. (4b)] and atrophy of remaining muscle fibers that is accompanied by an increased infiltration of non-contractile tissue (collagen, fat) [Fig. 4(a)]. For some muscles the reduction in muscle fiber number may start at a relatively early age (rectus abdominis muscle) (Inokuchi et al., 1975) whereas for other muscles the decline seems to accelerate from the sixth decade (Lexell et al., 1983, 1988) [Fig. 4(b)]. Single muscle fiber atrophy is observed at old age, where the type II muscle fibers are more affected than type I fibers (25–60% vs 0–25% reduced single fiber CSA) (Larsson et al., 1978; Klitgaard et al., 1990; Fiatarone Singh et al., 1999; Andersen 2003; Kosek et al., 2006) although reports of similar decrements in types I and II area also exist (Essén-Gustavsson & Borges 1986). At least, in part, the muscle fiber atrophy is caused by a reduced rate of myofibrillar muscle protein synthesis at old age (Balagopal et al., 1997; Welle & Thornton 1997).

Although not a universal finding (Hikida et al., 1998; Roth et al., 2000; Carlson et al., 2009), old individuals may show reduced amounts of myogenic satellite cells (Kadi et al., 2004; Sajko et al., 2004) that could contribute to the age-related loss in muscle fiber size. Regardless, load-mediated satellite cell activation has been reported to be reduced in aged compared with young rats (Gallegly et al., 2004). Similar results recently were obtained in aging individuals who showed impaired myogenic satellite cell activation due to diminished Notch signaling in response to short-term (14-day) lower limb immobilization followed by a period of strength training (Carlson et al., 2009).

In animal models (rat) the muscle atrophy observed during aging may not involve upregulation of enzymes involved in cellular protein degradation via the ubiquitin proteasome pathway (Edström et al., 2007). The same authors also showed that MuRFbx and Atrogin-1, which drive ubiquitin proteasome-mediated myofibrillar proteolysis, are downregulated in aged sarcopenic muscle, suggesting that the age-related sarcopenia may differ from the muscle atrophy occurring in disease or disuse (Edström et al., 2006). In contrast, elevated MuRF-1 and Atrogin-1 mRNA levels were observed in very old animals (Clavel et al., 2006). MuRF-1 but not Atrogin-1 were elevated in very old humans (85 years) (Raue et al., 2007) and did not differ between less old (72 years) and young individuals (Whitman et al., 2005). Collectively, these data suggest that the age-related sarcopenia may differ from the muscle atrophy associated with disease or disuse, at least below a certain age (in humans ∼80 years).

Changes in muscle mechanical function with aging

The impairment in neuromuscular function and the loss of skeletal muscle mass associated with aging have significant consequences for mechanical muscle function. Thus, marked decreases in maximum muscle strength, power, and RFD (measured as the rate of force rise: ΔFt) are observed with aging (Skelton et al., 1994; Izquierdo et al., 1999; McNeil et al., 2007; Caserotti et al., 2008b), even when chronically strength-trained elderly individuals are compared across the age range (Pearson et al., 2002). The loss in muscle mass [i.e. Fig. 4(a)] and the resulting reduction in mechanical muscle output leads to a loss in function (Janssen et al., 2002; Buchman et al., 2007) during typical activities of daily living such as rising from a chair, stair walking, and in postural balance control (Tinetti et al., 1988; Bassey et al., 1992; Skelton et al., 1994; Izquierdo et al., 1999; Foldvari et al., 2000). In further support of a close linkage between mechanical muscle function and functional capacity in the elderly, parallel and inter-related gains in functional performance and maximal muscle strength/power/RFD have been observed following strength training in old individuals (Fiatarone et al., 1994; Chandler et al., 1998; Suetta et al., 2004a; Caserotti et al., 2008a, b).

The capacity for rapid muscle force production can be evaluated by analysis of the contractile RFD (Aagaard et al., 2002, 2007). RFD is markedly reduced with aging, both when expressed in absolute terms or when normalized to body mass, respectively (Izquierdo et al., 1999; Barry et al., 2005; Klass et al., 2008). Notably, relative RFD (absolute RFD normalized to maximal isometric or dynamic strength capacity) also appears to be reduced with increased age (Izquierdo et al., 1999; Klass et al., 2008), suggesting that aging involves qualitative changes in skeletal muscles and/or in the neuronal system. Similar to muscle power, maximal RFD appears to decline with age even in highly sprint-/strength-trained individuals (Korhonen et al., 2006). Despite this decline, however, aged sprinters demonstrated markedly higher RFD than age-matched untrained elderly individuals.

Effects of strength training in the elderly

Physical exercise contains many different components that can be used to stimulate aerobic fitness, muscle strength and power, muscle mass, fine motor control, etc. Exercise may be characterized as endurance (aerobic), or resistance (strength) training, or a mix thereof. Endurance training consists of repeated low-force contractions with low-frequency muscle fiber activation patterns performed for a prolonged period (>20 min). In contrast, strength training exercise involves relatively few high-force contractions with high-frequency muscle fiber activation pattern applied intermittently (<2–4 min of total work per muscle group). Although the relative importance of these exercise components has not been examined in depth, strength training appears to offer some benefits over endurance training for improving neuromuscular function and increase muscle mass.

Adaptive changes in maximal muscle strength, power, and RFD

Although maximal muscle strength, power, and RFD decline with aging even in chronically (strength-) trained individuals (Pearson et al., 2002; Korhonen et al., 2006), significant beneficial effects of physical exercise may still be achieved. Thus, the capacity for rapid muscle force production (RFD) observed in elderly strength-trained individuals (Korhonen et al., 2006) is substantially greater (∼fourfold) than that reported in sedentary age-matched subjects (Izquierdo et al., 1999). Moreover, 80-year-old strength-trained lifters demonstrated similar maximal muscle power as that produced by 60-year-old untrained elderly individuals, suggesting that within this age range chronic strength training can effectively compensate for about 20 years of sarcopenic decline in muscle power (Pearson et al., 2002; their fig. 3). A substantial deficit (43%) in rapid force capacity (RFD) was observed between 80- and 60-year-old untrained women that was abolished in response to 12 weeks of explosive-type heavy-resistance strength training (Caserotti et al., 2008b) further indicating that a marked juvenilization in terms of mechanical muscle function (∼20 years) may be achieved when old individuals are engaged in strength training.

Maximal muscle power is found to increase after strength training in the elderly (De Vos et al., 2005; Caserotti et al., 2008a), including very old individuals (80 years) (Fielding et al., 2002; Caserotti et al., 2008b). Notably, greater gains in maximal muscle strength and muscular endurance were observed after strength training using heavy loads (80% 1RM) compared with less heavy loads (50% 1RM) (De Vos et al., 2005). Similarly, a large number of studies have demonstrated marked gains in maximal muscle strength of elderly individuals following heavy-resistance training (i.e. Häkkinen et al., 1998a, b, 2001; Hortobagyi et al., 2001; Suetta et al., 2004a; Barry et al., 2005; Caserotti et al., 2008b).

The capacity for rapid force production (RFD) and the contractile impulse (∫Fdt) is also increased after strength training in elderly individuals (Häkkinen et al., 1998a, 2001; Hortobagyi et al., 2001; Suetta et al., 2004a; Barry et al., 2005; Caserotti et al., 2008b), along with signs of elevated efferent neuromuscular activity as reflected by increased surface EMG amplitude (Häkkinen et al., 1998a 2001; Suetta et al., 2004a; Barry et al., 2005) (Fig. 7). Recent studies have demonstrated moderate to strong relationships both in young and old individuals between the integrated EMG amplitude and contractile RFD in the early phase of rising muscle force (Del Balso & Cafarelli 2007, Klass et al., 2008), suggesting that RFD is highly influenced by the magnitude of neuromuscular activity irrespective of age.

Figure 7.

 Explosive force capacity (rate of force development: RFD) in young (21–35 years, n=8) and old (60–79 years, n=8) individuals before (open bars) and after (hatched bars) 4 weeks of strength training using combined isometric elbow flexion and supination MVC. Training led to increased RFD in both young and old individuals (middle panels; *P<0.05), which was accompanied by increases in elbow flexor electromyography amplitude (bottom panels; *P<0.05) (data from Barry et al., 2005).

Adaptive changes in neuromuscular function

Strength training appears to elicit significant changes in neuromuscular function in elderly individuals. For example, aging individuals demonstrate elevated muscle EMG amplitudes during MVC contraction following a period of strength training, suggesting an increase in the magnitude of efferent neuromuscular activity (Häkkinen et al., 1998a, b, 2001; Suetta et al., 2004a; Barry et al., 2005).

Maximum RFD is influenced by maximal MN firing frequency (Baldissera et al., 1987; Garland & Griffin 1999; Van Cutsem & Duchateau 2005), and by the presence of MN discharge doublets (≤5 ms interspike interval) (Van Cutsem & Duchateau 2005). Notably, maximum MN firing frequency during isometric or dynamic ballistic MVC is reduced in old compared with young subjects (Connelly et al., 1999; Kamen & Knight 2004; Klass et al., 2008). Likewise, the incidence of MN doublet firing seems reduced in the elderly, during both rapid dynamic muscle contraction (Klass et al., 2008) and graded isometric ramp contractions (Christie & Kamen 2006). Importantly, strength training can increase maximum MN firing frequency in the elderly and fully eliminate the age-related difference in maximum MN firing frequency observed in the untrained state (Kamen & Knight 2004). Thus, MN firing frequency recorded in the VL muscle during maximal isometric quadriceps contraction increased in elderly (mean age 77 years, range 67–81 years) and young subjects (21 years, range 18–29 years) from 17.5 to 26 Hz and from 24 to 28 Hz, respectively, with no difference observed between the two age groups after the period of training (Kamen & Knight 2004). A fivefold elevated incidence of doublet discharge firing was reported after ballistic strength training in young subjects, which was suggested to be of importance for the concurrent gain in rapid muscle strength (RFD) (Van Cutsem et al., 1998). Future studies should clarify if similar effects can be achieved with strength training in aged individuals.

Elderly individuals may show reduced central muscle activation (CA) as assessed by electrical muscle stimulation superimposed onto MVC, albeit not a universal finding (for a review, see Klass et al., 2007). CA may increase with strength training in the elderly (Scaglioni et al., 2002; Morse et al., 2007; Suetta et al., 2007, 2009), and strong positive relationships (r=0.86–0.92) between individual changes in CA assessed by single-stimuli-interpolated twitch analysis in the quadriceps muscle and maximal isometric knee extensor strength (MVC) have been reported following strength training in very old individuals (80+ years) (Harridge et al., 1999) (Fig. 8). Unilateral long-term limb disuse in elderly subjects (∼70 years) due to hip arthrosis was accompanied by a marked reduction in CA for the affected limb (57.6%, 100%=full activation) vs the non-affected limb (CA: 70.7%) (Suetta et al., 2007). CA was reduced from 88.6% to 80.2% following short-term (2 weeks) limb immobilization in physically active elderly subjects (67.3 years, range 61–74 years) but remained unchanged (91.6% vs 90.6%) in young subjects (24.4 years, range 21–27 years) using interpolated twitch-doublet stimulation in the quadriceps muscle, indicating that the neuromotoric system of lower-limb muscles in old individuals may be more affected by short-term unloading than in young subjects (Suetta et al., 2009). Subsequent re-training by means of strength training (4 weeks) fully restored CA in old subjects (from 80.2% to 90.6%) whereas CA increased above pre-training levels in young subjects (from 90.6% to 95.2%), suggesting that the range of neuromuscular plasticity to retraining following immobilization may differ between old and young individuals (Suetta et al., 2009).

Figure 8.

 Central activation (CA) and maximal isometric knee extensor strength in very old individuals (85+ years, range 85–97, n=11) before and after a period of heavy-resistance strength training. CA was assessed using superimposed twitch interpolation (top panels) where superimposed twitches were greatly reduced after training along with marked increases in MVC strength (top panel showing illustrative individual). Relative gains in CA with training were strongly associated with the gain in MVC (bottom panels), also when MVC was normalized to knee extensor muscle CSA (right bottom panel) (data from Harridge et al., 1999).

The amount of antagonist muscle coactivation varies across contraction tasks. However, elderly individuals may show elevated antagonist muscle coactivation during leg extensor and arm flexor MVCs (Häkkinen et al., 1998a, 2001; Klein et al., 2001; Macaluso et al., 2002), although not a consistent finding (Häkkinen et al., 2000; Barry et al., 2005; Simoneau et al., 2005; Morse et al., 2007; Klass et al., 2008). If elevated before training, antagonist coactivation typically decreases in elderly individuals following strength training (Häkkinen et al., 1998a, 2001), although increased antagonist coactivation may also occur (De Boer et al., 2007). Thus, in old individuals where antagonist coactivation is elevated above normal levels (>20–25% maximum agonist activity) it is a typical finding that coactivation is decreased in response to strength training. Nevertheless, elderly individuals typically show elevated muscle coactivation during daily movement tasks such as during stair climbing and in single-step descent (Larsen et al., 2008; Hortobagyi & Devita 2000), which seems to be unaffected by strength training (Larsen et al., data in publication).

Fine motor control (force steadiness and force accuracy) appear to be impaired in the elderly, as indicated by an elevated variability (increased SD) and reduced matching accuracy in the muscle force (moment) produced during isometric and dynamic force tracking tasks (Hortobagyi et al., 2001; Tracy & Enoka 2002). Notably, fine motor control can be improved by strength training in the elderly, as evidenced by improvements in both force steadiness and force accuracy (Hortobagyi et al., 2001; Tracy et al., 2004; Tracy & Enoka 2006).

Adaptive changes in muscle size and architecture

Strength training using heavy loads (>70% 1RM, 10–14 weeks) leads to gains (5–12%) in muscle CSA and volume in elderly individuals, as evaluated by MRI or CT scanning (Frontera et al., 1988; Häkkinen et al., 1998a, b; Esmarck et al., 2001; Ferri et al., 2003; Reeves et al., 2004a; Suetta et al., 2004a). Notably, old and young individuals may demonstrate similar adaptive plasticity to strength training of moderate duration (3–5 months). Thus, reviewing a number of studies the relative gain in anatomical muscle size was not different between old and young individuals, respectively (0.07–0.12%/day vs 0.04–0.11%/day) (Narici et al., 2004, their Table 1).

Training-induced increases (20–40%) in single muscle fiber area have been observed when assessed in elderly individuals by muscle biopsy sampling techniques (Frontera et al., 1988; Häkkinen et al., 1998b, 2001; Hikida et al., 2000; Esmarck et al., 2001; Kosek et al., 2006; Suetta et al., 2008) including very old individuals (80+ years) (Kryger & Andersen 2007). However, the response may be blunted in the very old (≥80 years) compared with that seen in less old (<65 years) and young individuals (Fiatarone Singh et al., 1999; Martel et al., 2006; Slivka et al., 2008; Raue et al., 2009) suggesting that a limited adaptive plasticity in skeletal muscle growth may exist for the very old.

Training and maintained physical activity may not per se prevent the loss in type II fiber size with aging (discussed above). Thus, a trend for a decline in type II muscle fiber area was reported in elderly chronically trained sprinters compared with young sprinters, although fiber size was substantially greater in elderly sprinters than in untrained individuals of similar age (Korhonen et al., 2006). Similarly, old track and field athletes (sprint, jumping, throwers; 73.9 years, range 68–78) exposed to lifelong strength training showed 18–27% elevated type II muscle fiber area (with no difference in type I and type II fiber area) when compared with non-trained age-matched individuals, where muscle fiber area differed in the order IIX<IIA<I (Aagaard et al., 2007). Collectively, these data suggest that the age-related loss in muscle fiber size to some extent can be compensated by long-term (life-long) training involving strength conditioning exercise.

Similar to that seen in young individuals (Aagaard et al., 2001), adaptive changes in muscle architecture can be observed in elderly subjects in response to heavy-resistance strength training. Thus, following 12–52 weeks of training muscle fiber pennation angle increased in the resting VL muscle from 7.2 to 8.6° in frail elderly hip replacement patients (Suetta et al., 2008), from 11.3 to 14.5° in physically active elderly individuals (Reeves et al., 2004b), and from 15.4° to 17.3° in the lateral gastrocnemius, the latter measured during isometric MVC (Morse et al., 2007). A blunted adaptive change in muscle pennation angle was observed in previously physically active elderly compared with young subjects when 14 days of limb immobilization was followed by short-term strength training (4 weeks) (old: 9.0→8.4→8.6°; young: 10.4→9.4→10.5°), suggesting that training-induced changes in muscle architecture may take longer time to occur in older than younger individuals (Suetta et al., 2009). Importantly, the training-induced increase in fiber pennation angle allows for greater relative changes in single muscle fiber CSA than anatomical CSA (Aagaard et al., 2001; Suetta et al., 2008 vs 2004a), hence resulting in a qualitative improvement of the aging skeletal muscle that per se contributes to the gain in maximal muscle strength, RFD, and power.

Consequences for functional capacity

The improvement in mechanical muscle function induced by strength training in aging individuals frequently results in an improved functional capacity during tasks of daily living, especially in frail elderly or very old individuals (Fiatarone et al., 1994; Suetta et al., 2004a,b; Beyer et al., 2007; Caserotti et al., 2008b). Thus, in a classical study Fiatarone et al. (1994) demonstrated that strength training in very old (87.1 years) frail nursing home residents led to a 28% increased stair walking speed along with a 12% gain in maximal horizontal walking speed. Similar findings were reported in elderly post-operative hip replacement patients, where maximal horizontal walking speed, time for five repeated sit-to-stand movements, and maximal stair climbing speed were 28–30% improved following 12 weeks strength training (Suetta et al., 2004b). The increase in maximal gait speed was associated (r=0.79) with the improvement in rapid force capacity (RFD) (Suetta et al., 2004a), indicating that training-induced gains in mechanical muscle function can be a highly effective way to increase functional capacity in frail elderly individuals. In comparison, traditional rehabilitation training led to statistically non-significant changes of 0–19% (Suetta et al., 2004b).

Somewhat smaller, albeit statistically significant, improvement were observed following 36-week multicomponent training including strength exercises combined with balance, aerobic, flexibility, and coordination components in elderly males (75 years), which resulted in a 9–16% improved performance in chair rise test, 10 and 30 m maximal walking speed (Caserotti et al., 2008a). Similar findings have been observed in elderly women (78 years, range 70–90) who had experienced previous falls: 6 months of multicomponent strength training resulted in 10–21% improvement in sit-to-stand test performance and speed of maximal horizontal walking, and in stair climbing, respectively (Beyer et al., 2007). Importantly, the improvement in functional capacity remained present at follow-up testing 6 months after cessation of training (Beyer et al., 2007).

Methodological constraints

In the first half of this review article a large number of cross-sectional studies were reported. Obviously, the cross-sectional nature of these data inherently restraints the conclusions that can be made about the longitudinal physiological adaptation to aging. On the other hand, while very difficult to perform over a more extended age range (30–50 years) long-term longitudinal aging studies also to some extent are biased by selection (survival) of certain genetic subpopulations and/or lifestyle preferences, respectively. Consequently, valuable information can be obtained from cross-sectional studies, which subsequently may be scrutinized and explored in longitudinal intervention studies as described in detail in the latter half of the present review.


In conclusion, aging is characterized by loss of spinal MNs due to apoptosis, reduced IGF-1 signaling, elevated amounts of circulating cytokines, and increased cell oxidative stress. The age-related loss of spinal MNs is paralleled by a reduction in muscle fiber number and size (sarcopenia), consequently leading to an impaired mechanical muscle performance and a reduced functional capacity during everyday tasks. It is suggested that sarcopenic muscle may not per se be in a state of poor adaptive responsiveness, rather an impaired capacity for axonal reinnervation of deinnervated myofibers may be responsible for the net loss of muscle mass with advancing age. Concurrent decreases in maximum muscle strength, power, and rate of force development are observed with aging, even in highly trained master athletes. The impairment in muscle mechanical function is accompanied and partly caused by an age-related loss in neuromuscular function that comprise changes in maximal MN firing frequency, agonist muscle activation, antagonist muscle coactivation, force steadiness, and spinal inhibitory circuitry. However, elderly individuals demonstrate substantial adaptive plasticity in both skeletal muscles and the neuromuscular system in response to strength training (resistance exercise), which to a large extent can compensate for the age-related declines in muscle size and neuronal function, respectively, and lead to improved functional capacity even at very old age.