Relationship of anabolic hormones with motor unit characteristics in quadriceps muscle in healthy and frail ageing men.

CONTEXT
Anabolic hormones are important factors in maintaining muscle mass for ageing men, but their role in overall motor unit structure and function is unclear.


OBJECTIVE
To determine associations of anabolic and reproductive hormone levels with motor unit characteristics in quadriceps muscle in older healthy and frail men.


DESIGN
Observational cohort study of community dwelling men.


PARTICIPANTS
Healthy and frail men > 65 years old.


INTERVENTION
None.


OUTCOME MEASURE
Quantitative assessments of electromyography-derived motor unit potential size (MUP) and compound muscle action potential size (CMAP) of vastus lateralis muscle.


RESULTS
We studied 98 men (mean±SD: age 73±6 years; BMI 25.7±4.0 kg/m2; diabetes 11%) of whom 45% were prefrail and 18% frail. After adjusting for age, BMI and prevalent diabetes, higher total and free testosterone levels were significantly related to larger CMAP (total testosterone: β (95% CI): 0.3 (0.08, 0.53); free testosterone: 0.34 (0.13, 0.56)). Exploratory analysis showed the relationship between free testosterone and CMAP was stronger in frail rather than robust men. In univariate analyses, estradiol was associated with CMAP size (0.37 (0.16, 0.57)); and vitamin D was associated with MUP size (0.22 (0.01, 0.43)) but these relationships were no longer significant after adjusting for potential confounders.


CONCLUSION
Our data highlight the associations between androgen levels and the electrophysiological characteristics of older men, particularly in the frail. Clinical trials involving administration of androgens will help to elucidate the potential benefits of intervention on neuromuscular function and/or frailty status.


87
The declining numbers of motor units with advancing age(9,10) may cause 88 denervation of muscle fibers and constrain the ability of the central nervous system to 89 control voluntary movements. By way of compensation to preserve muscle function, some 90 denervated fibers can be reinnervated by axonal branching from neighbouring motor 91 neurons(11). This remodelling process leads to an increase in the size of surviving motor 92 units in older adults compared with young, but contributes to fiber atrophy and fiber 93 losses when reinnervation fails(12)(13). The underlying regulation of motor unit 94 remodelling in sarcopenia and frailty remains poorly understood, but may be associated 95 with hormonal changes during ageing, particularly declines in anabolic hormone levels.

96
The neuromuscular protective effects of androgens have been studied in animal 97 models in which male castration led to motor unit dendrites' atrophy, which was reversed 98 by testosterone administration (14,15). Similarly, when compared with controls, 99 testosterone therapy attenuated atrophy of motor neuron dendrites and muscle fibers in 100 female rats with spinal cord injury(16). Exogenous testosterone accelerated regeneration 101 of facial(17) and sciatic nerves(18) post injury and in humans, testosterone treatment 102 protected neuron cultures from cell death caused by testosterone deprivation(19). More 4 recent studies, however, suggest that dihydrotestosterone might be a more potent 104 anabolic hormone in mammalian skeletal muscle, exerting effects on force in slow and 105 fast twitch fibres alike(20). Dehydroepiandrosterone sulphate (DHEA-S) is a weak 106 androgen with neuroprotective and anti-apoptotic properties, which are independent of 107 any anabolic effects exerted after conversion to testosterone. Both in vivo and in vitro 108 models suggest that DHEA-S promotes neurogenesis, neuronal survival, and prevents 109 neurotoxicity due to its anti-glucocorticoid effects(21). None of these properties, 110 however, have been studied in the context of the peripheral nervous system and motor 111 neuron preservation in humans.

112
Similarly, cumulative evidence indicates that estradiol has a neuroprotective role 113 in the central nervous system(22). However, animal and human research also suggests 114 neuroprotective effects of estradiol on spinal motoneurons, where the Akt anti-apoptotic 115 signalling pathway is regulated by estradiol(23,24).

116
The anabolic role of vitamin D in the muscle has previously been studied in 117 health, sarcopenia and frailty predominantly in the context of muscle protein 118 turnover (25,26). Despite the fact that low levels of vitamin D have been linked to 119 impaired balance and frequent falls, no studies to date have investigated whether vitamin 120 D levels are related to neuronal control of muscle function and motor unit health.

121
Given the possible roles of these anabolic factors for human neuromuscular 122 function and the lack of data at the whole motor unit level in healthy ageing and frailty, 123 we aimed to determine the association between anabolic hormone levels and motor unit 124 characteristics in quadriceps muscles in older men from the general population.

158
Participants were invited to squeeze the handle as hard as possible for around 3 seconds 159 and the maximum contraction force (in kg) was recorded. This was repeated 2 times for 160 each hand, alternating between the right and left with 30 seconds rest between trials.   Table 1, alongside   172 the original CHS criteria(27). Individuals with one or more of these criteria were classed 173 as frail and those with none were classed as robust.

174
The FI is comprised of 37 health deficits (symptoms and signs, functional impairments), 175 which are known to accumulate with age and are associated with adverse health 176 outcomes. The FI was created using a standardized procedure(31). Continuous variables 177 were dichotomized based on the distribution of participants' scores; cut-off points were 178 set at the worst performing 10 th centile. Individuals with over 20% of missing data on 179 relevant deficits were excluded from the analysis. The details of the variables used to 180 create an FI and specific cut-off points are described in Supplemental

217
Linear regression models determined relationships between predictors 218 (hormone level) and outcome (MUP or CMAP). Each predictor as well as CMAP was 219 considered as an untransformed value standardized as a Z score [(raw score -220 mean)/standard deviation] to allow comparison of results between predictors. Motor 221 unit potential area, in view of significant skewing, was log-transformed before being 222 standardized as Z score to meet the linear regression assumptions. 223 Models were adjusted for age, body mass index (BMI), diabetes and alcohol excess 224 as these correlated with the predictors and therefore were potential confounders. The 225 analyses where estradiol was a predictor were further adjusted for total testosterone -226 the main precursor of E2 production in men. The results of these analyses were displayed 227 as standardized coefficients (beta) with 95% confidence intervals.

228
In an exploratory analysis, we introduced an interaction term (hormone x frailty 229 phenotype or hormone x frailty index) as well as a FP or FI variable, as appropriate, to the 230 fully adjusted models to assess whether the relationships between hormone levels and 231 EMG parameter values varies in health in relation to the level of frailty.

244
These data indicated that age, BMI, diabetes and alcohol excess might be potential 245 confounders and therefore we included these as covariates in subsequent models.

246
In unadjusted analysis T, free T, DHT and E2 were positively related to CMAP size:  (Table 3).

252
In unadjusted analysis one SD higher level of free T and vitamin D was associated 253 with larger mean MUP. However, these associations were no longer statistically 254 significant after BMI adjustment (Table 3).

258
The relationship between free T and CMAP was greater with increasing frailty

264
When we explored associations between physical function and EMG parameters, 265 Timed up and go (TUG) was negatively related to MUP size, and the association was 266 partially attenuated after adjusting for lean muscle mass in keeping with a partial 267 mediation model (Supplemental Table 3)(27). TUG was negatively linked with CMAP and 268 the associations appeared largely independent of lean muscle mass (Supplemental Table   269 3)(27). Knee extensor maximum voluntary contraction (KEMVC) was associated with 270 MUP size and this association appeared to be explained by lean muscle mass 271 (Supplemental Table 3)(27). We did not observe a significant relationship between knee 272 extensor MVC and CMAP.

273
Finally, we performed an analysis assessing the potential influence of selection 274 bias. Compared to the 16 men (14%) who were excluded because of incomplete data, the 9 98 men in the study cohort were less likely to be frail and were less likely to have 276 sarcopenia, weakness, respiratory disease, diabetes and arthritis (Supplemental Table   277 4)(27). Therefore, the strength of relationships described above may be conservative in unadjusted models and also models adjusted for BMI, age and prevalent diabetes. We 286 have also observed a similar trend for DHT, however adjustment for diabetes attenuated 287 the DHT-CMAP relationship. Secondly, we showed that estradiol was related to muscle 288 CMAP, but this relationship was rendered non-significant after adjusting for total 289 testosterone levels indicating that the effect is likely to be testosterone-related. Thirdly, 290 we showed that vitamin D was positively related to motor unit potential size in 291 unadjusted models, but this effect was no longer significant after adjusting for BMI.

338
We showed that low free testosterone, which is a biologically active fraction of 339 circulating testosterone, is associated with impaired muscle electrophysiology assessed 340 by maximal CMAP. The maximal CMAP size, whilst not dependant on total size of larger 341 muscle groups (7)

377
Whereas some of the effects of vitamin D deficiency on muscle are thought to be, 378 in part, mediated by raised pro-inflammatory cytokines levels(26,56) and direct 379 activation of the IGF-1 receptor(57,58), vitamin D deficiency has previously been linked 12 to altered muscle innervation(59) which is further supported by our findings .

383
Our study has a number of strengths. Our cohort is representative of older community-384 dwelling men and although the sample size may appear small, it is relatively large for an 385 invasive study in elderly and frail participants. To our knowledge, it is the first study in 386 humans to relate hormone levels to motor unit size and muscle electrophysiological 387 characteristics assessed by intramuscular and surface electromyography. However, we 388 did not measure calcium or PTH levels which may have helped in the interpretation of the 389 vitamin D data. We have also not performed nerve conduction studies, which could have 390 helped in the interpretation of study findings to identify whether deficits exist in motor 391 neuron axons. Our work was limited to men, so the generalizability to women is unknown.