The contribution of age and sex hormones to female neuromuscular function across the adult lifespan.

S.O.B. was involved in conception and experimental design, acquisition, analysis and interpretation of data, and drafting and revising the manuscript for intellectual content. A.C. was involved in acquisition, analysis and interpretation of data, and drafting and revising the manuscript for intellectual content. C.F. was involved in analysis of data and revising the manuscript for intellectual content. D.H. was involved in conception and experimental design, acquisition, analysis and interpretation of data and drafting and revising the manuscript for intellectual content. S.L. was involved in conception and experimental design, interpretation of data and drafting and revising the manuscript for intellectual content. All authors approved the final version to be published and agree to be accountable for all aspects of the work.

S.L. and this study were supported by an Australian Research Council Future Fellowship (FT10100278).

The present study was approved by Deakin University's Human Research Ethics Committee (DUHREC 2021‐307) and conducted in accordance with the standards set by the Declaration of Helsinki , except for registration in a database. All volunteers provided written informed consent after being provided with a digital copy of the plain language summary of the study, which detailed the experimental procedures, associated risks and the liberty to withdraw consent at any time without jeopardy. Eligibility for inclusion included apparently healthy biological females between 18 and 80 years of age. Health status was determined via a medical history questionnaire and exclusion criteria included pregnancy, recent history of cancer, implanted medical devices, body mass index (BMI) > 35 kg/m 2 or musculoskeletal injury/pain of the tibiofemoral or patellofemoral joint complex. Participants were also ineligible if they used medications that are known to impact skeletal muscle metabolism, such as corticosteroids, thyroxine, metformin or semaglutide. Participants were instructed to avoid strenuous physical activity 24 h before and on the day of neuromuscular assessments and to avoid caffeine on the day. Eighty‐eight biological female participants completed the study, with all decades of adulthood represented in the cohort (18–29 years: n  = 19, 30–39 years: n  = 13, 40–49 years: n  = 12, 50–59 years: n  = 16, 60–69 years: n  = 15, 70–80 years: n  = 13). Participant characteristics including measures of body composition, neuromuscular function, reproductive health and lifestyle components are shown in Table  1 . Females were recruited at every stage of the reproductive span (premenopausal: n  = 43, perimenopausal: n  = 3, postmenopausal: n  = 42). In total, 41% of pre‐ or perimenopausal participants used a form of hormonal contraception [intrauterine device (IUD), oral contraceptive pill (OCP) or implant], and 18% of peri‐ or postmenopausal participants used hormonal replacement therapy. Participant characteristics. Data are presented as mean (SD); n / N (%). Abbreviations: ALM, appendicular lean mass; BMI, body mass index; BW, body weight; CSA, cross‐sectional area; H MAX . M MAX , highest peak‐to‐peak amplitude of the H‐reflex normalized to the peak‐to‐peak amplitude of the maximal compound muscle action potential; HRT, hormonal replacement therapy; IUD, intrauterine device; M DUR , quadriceps maximal compound muscle action potential peak‐to‐peak duration; M MAX , quadriceps maximal compound muscle action potential peak‐to‐peak amplitude; e1RM, estimated leg press one‐repetition maximum; MVC, maximal voluntary contraction; MVPA, moderate ‐to‐vigorous physical activity; OCP, oral contraceptive pill; PT, peak potentiated resting twitch force; RMS. M MAX , quadriceps root mean square normalized to maximal compound muscle action potential; RTD, rate of torque development; VA, voluntary activation. Specific information pertaining to quadriceps neuromuscular assessment methodology including electrical stimulation, surface electromyography, and data acquisition and analysis have all been described in detail by our group previously (O'Bryan et al., 2024 ). Briefly, isometric neuromuscular assessments were conducted on an isokinetic dynamometer (Universal Pro Single Chair model 850‐230, Biodex Medical Systems, New York, NY, USA). Participants sat upright in the dynamometer chair with hip and knee angle set at 85° and 75° flexion, respectively. Participants began each neuromuscular assessment with a standardized warm‐up procedure including a series of voluntary submaximal (20, 40, 60 and 80% perceived effort) and minimum of three maximal (100% perceived effort) isometric knee extensions and knee flexions on thier dominant leg (4 s push, 1 s relax, 4 s pull) until a plateau in voluntary maximal torque was recorded (60 s between maximal efforts). Following the warm‐up, participants performed an ~4 s maximal isometric voluntary contraction (MVC) of the knee extensors with an electrically evoked doublet (100 Hz) applied to the femoral nerve at the plateau in voluntary torque, followed ∼2 s after by three resting evoked twitch responses (100, 10 and 1 Hz ∼1.5 s apart). This procedure was repeated three times (2 min rest separating each set) with the highest recorded MVC (N·m), quadriceps root mean square normalized to maximal compound muscle action potential (RMS. M MAX ), voluntary activation (VA, %) peak potentiated resting twitch torque (100 and 10 Hz in N·m plus a 100:10 Hz ratio), maximal rate of torque development (RTD, N·m.s −1 ) and quadriceps maximal compound muscle action potential peak‐to‐peak amplitude ( M MAX, mV) and duration ( M DUR. s) calculated for analysis. After an ∼15 min rest period, Hoffman reflexes (H‐reflex) were elicited in quadriceps by applying a 1 Hz electrical stimulus to the femoral nerve at progressively increasing intensity during 50 brief (2—3 s) submaximal isometric contractions performed at 5% MVC (10 s rest separated contractions), with the highest peak‐to‐peak amplitude of the H‐reflex normalized to the peak‐to‐peak amplitude of the maximal compound muscle action potential ( H MAX . M MAX ) included for analysis. Finally, after another ∼15 min rest period, dynamic leg strength was assessed with a leg‐press five‐repetition maximum (RM) test. Participants completed up to four test sets of five repetitions. After each test set, participants were asked to rate their perceived exertion from 1 to 10 on the modified Borg Scale (Borg, 1982 ). The resistance was increased until the participant communicated a 10 rating (i.e. maximally exerted) after five repetitions. The corresponding weight was recorded as their 5RM (kg), and a modified equation was used to calculate estimated one‐repetition maximum leg press (e1RM in kg) (Wood et al., 2002 ): Estimated e 1 RM = weight / 1.013 − 0.0267123 ∗ repetitions The first 25 participants completed two neuromuscular assessment sessions (minus 5RM test) with excellent inter‐session reliability established for all isometric voluntary and evoked outcomes (O'Bryan et al., 2024 ). Accordingly, the remaining 63 participants were assessed once. Herein, voluntary and evoked isometric torques and e1RM were additionally reported relative to quadriceps lean muscle CSA (N·m.mm − 2 ) to describe muscle‐specific torque and quality (Pöllänen et al., 2011 ). Participants recorded their physical activity across a 7 day period with an actiCAL accelerometer watch (Respironics Inc., Murraysville, PA, USA). This watch records energy expenditure and intensity of physical activity and has been validated against oxygen consumption‐computed active energy expenditure (Heil, 2006 ). During the hours in which participants were awake (determined by a sleep diary) the time spent in moderate and vigorous physical activity (>1535 counts per minute; MVPA) per day was calculated and averaged across all valid days (Colley & Tremblay, 2011 ). Days were considered valid if there was >10 h of data available, and participants required at least four valid days of data to be included in the physical activity analysis. Eight participants did not meet these criteria and were excluded from physical activity analysis. During the same 7 day period, participants were asked to record all food and drink intake via the Easy Diet Diary app (Xyris Software, Brisbane, Australia). Average protein intake (g.kg body weight −1 .day −1 ) was calculated from participants with at least four full days of data. Twelve participants did not meet these criteria. Fasted plasma samples were taken from the antecubital vein, immediately centrifuged at 4°C, and stored at −80°C until analysed. Oestradiol, testosterone, progesterone, luteinizing hormone (LH) and follicle‐stimulating hormone (FSH) were measured by high performance liquid chromatography mass spectrometry (Triple Quad 5500, Sciex, Framingham, MA, USA) at Monash Health Pathology laboratory (Victoria, Australia). Conjointly, a menstrual cycle history questionnaire was used to determine self‐reported menopausal stage as well as use of hormonal contraception or menopausal hormone replacement therapy. Information was also collected about the last known menstrual cycle, reproductive conditions (polycystic ovary syndrome: n  = 5, endometriosis n  = 2, fibroids: n  = 5, or ovarian cysts: n  = 5) and symptoms of perimenopause or menopause (irregular periods, muscle/joint pain, sleep disturbances, hot flushes/night sweats, irritability, crawling feelings under the skin) where relevant. To determine menstrual cycle phase on the day of testing, premenopausal participants collected a urine sample on the morning of their neuromuscular testing session. Enzyme linked immunosorbent assays (ELISAs) specific to the urine matrix were used to quantify oestradiol (Abcam, Cambridge, UK) and progesterone (Invitrogen, Waltham, MA, USA). Urinary hormone levels were used in addition to a menstrual cycle calendar (e.g. day of cycle, last known cycle) to determine menstrual cycle phase using a modified version of Elliott‐Sale et al. ( 2021 ) to account for urinary hormone concentrations. Dual energy X‐ray absorptiometry (DEXA; GE Lunar, Madison, WI, USA) was used to assess body composition, including total body fat and lean mass (kg), and bone mineral density (BMD, mg.cm −3 ). During the scan participants lay in a supine anatomical position with the hands in a neutral position. A foam block was placed between the arms and trunk to separate the regions. The enCORE software (GE Healthcare, Chicago, IL, USA) uses skeletal landmarks in the image to detect regions of interest, which were manually adjusted where appropriate. The leg region was determined by placing a border through the femoral neck and the arm region was defined by a border at the medial side of the humerus neck. The sum of leg and arm lean mass was used to calculate appendicular lean mass (kg). A peripheral quantitative computed tomography (pQCT) scan (voxel size: 0.5 × 0.5 mm, scanning speed: 20 mm s −1 , XCT 3000, Stratec Medizintechnik GmBH, Pforzheim, Germany) of the upper thigh was obtained to determine quadriceps muscle CSA and intramuscular fat area. Participants lay in a supine position with their measured leg secured in a footrest. Tibial length was used as an approximation of femur length, measured from the tibial plateau to the medial malleolus while the knee was flexed to 90° (Cervinka et al., 2018 ). The images were taken at 50% of the tibia length from the mid‐condylar cleft towards the hip and analysed using ImageJ software (version 2.0.0; National Institutes of Health, Bethesda, MD, USA). Movement artefacts on the scan were rated from 1 (none) to 5 (extreme) as previously described (Blew et al., 2014 ). Images with a score ≥4 were excluded from analysis ( n  = 8). Lean muscle CSA and intramuscular fat of the quadriceps were determined by manual tracing regions of interest using ImageJ software, followed by automated thresholding to segment fat and muscle CSAs (Fuchs et al., 2023 ). Image pre‐processing included contrast enhancement, Gaussian blur, and threshold‐based quantification of fat and muscle areas. The same two researchers reviewed all images in a blinded manner to ensure consistency and accuracy in analysis. All data were analysed using Rstudio 4.3.2 (R Core Team, 2021 ). Missing data in predictors were imputed using the kNN function in the VIM package (Kowarik & Templ, 2016 ). To explore the relationship between age and neuromuscular measures adjusted for lifestyle variables (MVPA, protein intake), scatterplots were created to visualize the data distribution. For bounded variables (i.e. percentages), beta regression models of the form ( N e u r o m u s c u l a r M e a s u r e ∼ A g e + M V P A + P r o t e i n ) were run. Further examination of the scatter plots and residuals of linear models indicated that unbounded variables displayed a non‐linear relationship with age. Therefore, generalized additive modelling (GAM) analyses were run with thin plate regression splines and using the restricted maximum likelihood (REML) method to optimize smoothness (Wood, 2017 ). The model was of the form: N e u r o m u s c u l a r M e a s u r e ∼ s ( A g e ) + M V P A + P r o t e i n , m e t h o d = R E M L . Finally, we adjusted our isometric MVC and e1RM models to account for the CSA of lean muscle in the quadriceps. The model was of the form N e u r o m u s c u l a r M e a s u r e ∼ s ( A g e ) + M V P A + P r o t e i n + Q u a d _ C S A , m e t h o d = R E M L and expressed as a specific force measurement. Model fit was assessed through effective degrees of freedom (EDF) and the k ‐index ( k ′), with values closest to 1 indicating linearity and poor fit, respectively. Since the EDF values indicated a non‐linear change and visualization of the GAM models suggested a steepness of the slope around the menopausal transition, we identified where this change occurred using the first and second derivatives. The first derivative indicates the slope of the tangent line to the curve, and the second derivative highlights inflexion points. To pinpoint regions of steep decline in the outcome variable, we focused on where the first derivative dropped below 75% of its maximum value. We then selected the subset of the sample located beyond this threshold to conduct principal component analysis (PCA). The principal components (PCs) were identified and those with eigenvalue >1 (i.e. collectively explaining more of the variance than one predictor variable alone) were retained. Correlations between the predictor variables and PCs were visualized in a correlation matrix plot to understand the importance of each variable to the various PCs. We then conducted correlation analysis of the neuromuscular measures with the PCs to identify significant associations. Correlations ( r ) were classified as strong (| r | ≥ 0.7), moderate (0.7 < | r | ≥ 0.3), weak (| r | < 0.3) or negligible (| r | < 0.1) (Cohen, 1988 ). Other packages used in our analysis include FactoMineR (Lê et al., 2008 ), factoextra (Kassambara & Mundt, 2020 ), mgcv (Wood, 2017 ), lme4 (Bates et al., 2015 ), lmerTest (Kuznetsova et al., 2017 ) and tidyverse (Wickham et al., 2019 ).

This study is part of a larger study relying on the collection of muscle biopsies during the early follicular phase of the menstrual cycle in premenopausal females (and in perimenopausal females, where possible), or at any time in postmenopausal females. As a result, and while all efforts were made to collect neuromuscular data during the same menstrual cycle phase, this was not always possible. As a first step, we verified whether menstrual cycle phases had a significant effect on any of the neuromuscular outcomes. All measures but two (rectus femoris M MAX and vastus medialis M DUR , both P  < 0.05) displayed no significant association with menstrual cycle phase (Table  2 ). The same analysis was run to investigate the correlation between hormonal contraception and hormone replacement therapies with neuromuscular outcomes. There were no significant associations between hormone replacement therapy or hormonal contraception mode (none, implant, IUD or OCP) with any of the neuromuscular outcomes (Table  2 ). Effects of hormonal contraception, menstrual cycle and hormone replacement therapy on neuromuscular outcomes. Abbreviations: H MAX . M MAX , highest peak‐to‐peak amplitude of the H‐reflex normalized to the peak‐to‐peak amplitude of the maximal compound muscle action potential; HRT, hormonal replacement therapy; M DUR , quadriceps maximal compound muscle action potential peak‐to‐peak duration; M MAX , quadriceps maximal compound muscle action potential peak‐to‐peak amplitude; e1RM, estimated leg press one‐repetition maximum; MVC, maximal voluntary contraction; PT, peak potentiated resting twitch force; RMS. M MAX , quadriceps root mean square normalized to maximal compound muscle action potential; RTD, rate of torque development; SE, standard error; VA, voluntary activation. Statistical significance accepted as P  < 0.05 and indicated with an asterisk ( * ). Beta‐regression models were conducted for bounded neuromuscular outcomes (i.e. outcomes expressed as percentages) including voluntary activation and quadriceps H‐reflex and RMS. M MAX . None of these variables were significantly associated with age (Table  3 ). Impact of age on bounded neuromuscular outcomes using beta‐regression. Abbreviations: H MAX . M MAX , highest peak‐to‐peak amplitude of the H‐reflex normalized to the peak‐to‐peak amplitude of the maximal compound muscle action potential; RMS. M MAX , quadriceps root mean square normalized to maximal compound muscle action potential; SE, standard error; VA, voluntary activation. All analyses are adjusted for MVPA and protein intake. Statistical significance accepted as P  < 0.05. All other outcomes were modelled using GAMs. The smooth terms for MVC (N·m), e1RM (kg), RTD (N·m.s −1 ), PT 100 (N·m) and PT 10 (N·m) indicated a declining non‐linear relationship with age ( P  < 0.001; Fig.  1 A – E ), meaning that, as age increased, these outcomes decreased in a non‐linear manner. Rectus femoris M MAX (mV), quadriceps lean CSA (mm 2 ) and specific e1RM (kg.mm −2 ) were associated with age in a negative, quasi‐linear manner ( P  < 0.05, Fig.   1 G – I ). No other outcomes were significantly associated with age (Table  4 ) including specific MVC (N·m.mm −2 ) (Fig.  1 F ) and quadriceps intramuscular fat CSA (mm 2 )(Fig.  1 J ). All neuromuscular outcomes were modelled using generalized additive models (GAMs). The red shaded region highlights a rapid decline in the outcome, starting at the point where its first derivative falls below the 75% threshold. Each panel depicts the partial plot of age adjusted for MVPA and protein intake. edf = estimated degree of freedom. P  < 0.05 was considered statistically significant. Impact of age on neuromuscular outcomes using generalized additive models. Abbreviations: EDF, effective degrees of freedom; M DUR , quadriceps maximal compound muscle action potential peak‐to‐peak duration; M MAX , quadriceps maximal compound muscle action potential peak‐to‐peak amplitude; e1RM, estimated leg press one‐repetition maximum; MVC, maximal voluntary contraction; PT, peak potentiated resting twitch force; RTD, rate of torque development. All analyses are adjusted for MVPA and protein intake. Statistical significance accepted as P  < 0.05. The fitted GAMs that were significant for age and displayed a significant EDF value greater than 1 were selected for further analysis. The steeper decline of each curve, defined as the point where the first derivative fell below 75% of its maximum value, started at 42.9 years of age for PT 10 , 44 years of age for RTD, 45.2 years of age for PT 100 , 46.4 years of age for MVC and 47.6 years of age for e1RM (Fig.  1 A – E , red shaded region). A second inflexion for e1RM was also observed at 67.4 years. PCA was conducted on the subset of the sample aged beyond the relevant inflexion points to investigate the contribution of age, lifestyle factors (MVPA and protein intake), quadriceps tissue composition and sex hormones to the observed decline of neuromuscular outcomes postmenopause. Principal components 1 to 4 accounted for 69.2% of the total variance in the data (PC1 = 24.7%; PC2 = 18.2%; PC3 = 15.9%; PC4 = 10.4%). Each of these components had eigenvalues greater than 1, indicating that they explained more variance than any individual variable, and thus, were considered significant in capturing the underlying patterns of the dataset. The significant variables within each component differed between each PC (Fig.  2 A ). Correlation of the PCs with neuromuscular variables that were significantly affected by age [e1RM (kg), MVC (N·m), RTD (N·m.s −1 ), PT 10 (N·m) and PT 100 (N·m)] were then examined. PC1 was positively correlated with MVC ( r  = 0.37), PC3 was negatively correlated with PT 10 ( r  = −0.45) and PT 100 ( r  = −0.41), and PC4 was negatively correlated with MVC ( r  = −0.41), RTD ( r  = −0.56), PT 10 ( r  = −0.64) and PT 100 ( r  = −0.62) (Fig.  2 B ). e1RM was not significantly correlated with any of the PCs, but showed a weak to moderate positive correlation with PC1 ( r  = 0.31, P  = 0.051). A ) Correlogram: correlations between the predictor variables and the principal components. B ) Correlogram: correlations between neuromuscular outcomes and principal components. Negative correlations are shown in purple, and positive correlations are shown in green. * A significant correlation at P  < 0.05.

This study aimed to investigate the determinants of age‐related changes in quadriceps neuromuscular function across the adult female lifespan. Analysis of 88 healthy females represented equally across each decade of life from 18 to 80 years and adjusted for the lifestyle factors physical activity level and protein intake, demonstrated a non‐linear decline and accelerated reductions occurring around menopausal onset in all voluntary and evoked torque responses, whereas quadriceps lean CSA and rectus femoris compound muscle action potential amplitude decreased quasi‐linearly with age. No significant effect of age was observed for isometric MVC specific to quadriceps lean CSA, although e1RM specific to quadriceps lean CSA decreased. Outcomes representative of excitatory neural drive to quadriceps were unaffected by age, including voluntary activation, H‐reflexes and RMS normalized to maximal muscle compound action potential for vastus lateralis, vastus medialis and rectus femoris muscles. PCA revealed that, in postmenopausal females, age‐related reductions in neuromuscular variables could be explained by inter‐individual differences in quadriceps tissue composition, lifestyle factors and concentrations of sex hormones oestradiol, progesterone and testosterone. Non‐linear age‐related reductions in muscle torque starting in the fourth decade of the female lifespan aligns with previous observations of muscle strength (Lindle et al., 1997 ) and general molecular dysfunction within several physiological systems (Shen et al., 2024 ). Muscle torque is highest when shortening velocity is lowest (Thorstensson et al., 1976 ) and torque‐generating capacity is directly proportional to the CSA of a muscle (Jones et al., 2008 ). The lack of age‐related change in isometric voluntary torque specific to quadriceps lean CSA (both normalized and statistically adjusted) supports the considerable role that muscle size/hypertrophy plays in isometric torque production across the female lifespan. In the present study, quadriceps lean CSA decreased quasi‐linearly at an average rate of ∼28 mm 2 per year, slightly less than the linear decrease reported from 40 years of age for healthy females (Mizuno et al., 2021 ). However, the decline in isometric MVC was non‐linear with an inflexion point reported at 46.4 years of age, suggesting that quadriceps lean CSA becomes relatively more important for isometric torque production beyond this age. Maintenance of skeletal muscle mass later in life has vital implications for metabolic and musculoskeletal health and the prevention of age‐related diseases such as type 2 diabetes, sarcopenia and osteoporosis (Evans, 1997 ; Lombardi et al., 2016 ). Although the lack of change in isometric MVC relative to quadriceps lean mass aligns with some studies (Laakkonen et al., 2017 ), others have shown a decrease in isometric muscle‐specific torque with increasing age (Wrucke et al., 2024 ). The reason for this discrepancy is unclear but may be related to methods adopted to quantify quadriceps lean CSA. Here, we used computed tomography and extracted lean CSA from the quadriceps only and independent of non‐contractile tissue (intramuscular and subcutaneous), which is a validated technique with high level of agreement with gold standard magnetic resonance imaging (MRI) methods (Fuchs et al., 2023 ). Discordant with the physiological determinants of isometric torque with advancing age, quadriceps type II fibres preferentially atrophy in older age (Nilwik et al., 2013 ) and explain a large proportion of the age‐related decline in dynamic strength and power (Power et al., 2013 ). Moreover, loss of high‐threshold motor units in older adults decreases motor unit firing frequencies (Piasecki et al., 2016 ; Wages et al., 2024 ) and older females show less musculotendinous stiffness than their younger counterparts (Wu et al., 2016 ). Indeed, muscle size plays less of a role in e1RM compared to isometric MVC, as significant decreases across the lifespan remained evident when e1RM was both normalized to, and statistically adjusted for, quadriceps lean CSA. Although dynamic torque and muscle shortening velocity decrease earlier in the lifespan and at greater magnitude compared to torque production at slower speeds (Haynes et al., 2020 ; Raj et al., 2010 ), we did not observe any differences in the initial age of decline for isometric MVC (46.4 years) compared to dynamic e1RM (47.6 years). Quadriceps activation is similar during isometric knee extension and leg press exercise (Alkner et al., 2000 ) and knee extension joint power can effectively model torque–velocity relationships in leg press exercise (Bobbert, 2012 ), demonstrating the significant contribution of the quadriceps to leg press performance. However, the lack of difference in the initial age of decline for isometric MVC and dynamic leg press may have been confounded by contribution of other large muscle groups which also demonstrate near maximal activation levels during leg press exercise at heavy loads (e.g. gluteus maximus) (Da Silva et al., 2008 ) or from the potential influence of the muscle stretch–shortening cycle from a repetitive 5RM test (Cormie et al., 2011 ). Nonetheless, reductions in dynamic torque and power have greater negative consequences for overall functional capacity (Power et al., 2013 ) and are more prevalent in older females compared to males (Wrucke et al., 2024 ). Indeed, we did observe a second inflexion point in e1RM at 67.4 years of age which was not evident in isometric MVC, suggesting a second period of potential vulnerability to age‐related decline in dynamic strength and power. Interestingly, the initial inflexion point of accelerated age‐related reductions in MVC and e1RM (46 and 47 years) occurred after the observed reduction in PT 10 (42.9 years), RTD (44 years) and PT 100 (45 years). Although PT 10 seemed to reduce earlier in the lifespan than PT 100 , the lack of change in the high‐ to low‐frequency ratio suggests a somewhat comparable rate of impairments in different peripheral compartments. Considerable reductions in high‐ and low‐frequency evoked torque outcomes have previously been shown in comparisons between younger and older females (Solianik et al., 2017 ; Varesco et al., 2022 ; Wrucke et al., 2024 ), but here we illustrate the specific age at which marked reductions start to occur around menopause onset during the early‐to‐mid fourth decade. The reduced peak torque during high‐frequency stimulation may indicate neuromuscular junction instability arising from age‐related morphological alterations (e.g. number and affinity of acetylcholine receptors, reduced junctional folds and increased fragmentation) (Arnold & Clark, 2023 ; Hepple & Rice, 2016 ; Piasecki et al., 2016 ) or increased sarcolemma refractoriness related to sodium channel inactivation (Lee et al., 2018 ). Indeed, C‐terminal agrin fragment (CAF) as a biomarker of neuromuscular junction degradation is elevated in postmenopausal compared to premenopausal females (Willoughby et al., 2024 ). Reduced peak torque during low‐frequency stimulation suggests age‐related impairments in Ca 2+ handling/content (Jones, 1996 ; Millet et al., 2011 ) known to occur in elderly females (Hunter et al., 1999 ) and related to reduced maximal rate of torque development and strength of actin–myosin bound states in males (Mazara et al., 2021 ). The earlier onset of age‐related impairments in evoked compared to voluntary torque responses suggests that intrinsic muscle function may be impaired earlier in the lifespan than any observable decline in functional capacity, perhaps related to compensatory increases in motor unit firing frequencies resulting from reduced muscle quality. Although poor muscle quality is suggested to in part explain higher motor unit firing rates in older females compared to males (Guo et al., 2024 ), this requires further investigation within females and during high force contractions. Based on our analysis, quadriceps intramuscular fat data did not significantly change with age, although it tended to increase across the decades from 40 to 70 years in line with previous findings utilizing pQCT methods (Mizuno et al., 2021 ). Advanced MRI approaches could serve as a more effective method to accurately assess intramuscular fat content in very old adults, given its enhanced ability to distinguish muscle from fat. This approach has showed increased quadriceps intramuscular fat content in 70–80‐year‐old compared to 20–30‐year‐old females (Hogrel et al., 2015 ). The lack of an age effect on quadriceps voluntary activation, H‐reflex amplitude and RMS normalized to maximal muscle compound action potential suggests that steep reductions in quadriceps torque and neuromuscular function starting during the fourth decade and during the menopausal transition were largely mediated by peripheral muscular mechanisms occurring at or distal to the neuromuscular junction, and not an incapacity of supraspinal and spinal circuits to volitionally recruit existing quadriceps motor units or reduced facilitation of the motoneuron pool via Ia afferent stimulation. Previous studies have shown that compared to younger cohorts, quadriceps voluntary activation is lower in older females when assessed by train stimuli (Solianik et al., 2017 ), decreased in older mixed‐sex cohorts with high‐frequency paired pulse stimulation (Mau‐Moeller et al., 2013 ), or unchanged when assessed via transcranial magnetic stimulation (Wrucke et al., 2024 ) or magnetic nerve stimulation (Varesco et al., 2022 ). Here, we implemented a gold‐standard technique to assess voluntary activation of large functional quadriceps muscles in females via high‐frequency electrical paired pulse twitch interpolation (Nuzzo et al., 2019 ) and normalized voluntary EMG RMS amplitude to maximal M‐wave amplitude to account for muscle size and peripheral transmission confounding effects (Millet et al., 2011 ). Further, as a history of long‐term aerobic physical activity may promote re‐innervation of denervated fibres (Piasecki et al., 2019 ) without impacting muscle size or isometric and dynamic torque in older adults (Chambers et al., 2020 ), we adjusted out statistical models to account for MVPA (and protein intake). This was especially important as all participants in the present study including those above 65 years met (or exceeded for those 50–59 years) the minimum guidelines for moderate to vigorous physical activity (150–300 min per week) (Bull et al., 2020 ), reflecting a well‐known recruitment bias in exercise physiology studies (Stefanetti et al., 2014 ). After these adjustments, voluntary activation was not significantly associated with age, indicating that the heterogeneity in current findings is probably related to assessment methods including stimulation parameters (type, location of the anode and cathode, intensity), compliance and sensitivity of the myograph, timing of the stimulation and/or length of the muscle (Nuzzo et al., 2019 ; Rozand et al., 2020 ). Although MVPA did not influence quadriceps voluntary activation, older adults with a resistance strength training history may have higher levels of voluntary activation (Arnold & Bautmans, 2014 ) and this may need to be specifically assessed (e.g. Junior et al., 2021 ) and adjusted for when evaluating age‐related changes in the capacity to volitionally recruit existing motor units. Rectus femoris M‐wave amplitude decreased with increasing age without any observed changes for vastii muscles, suggesting potential vulnerability of this bi‐articular quadriceps muscle to age‐related impairments in neuromuscular function across the female lifespan. Rectus femoris typically demonstrates a higher percentage of MyHC type II fibre isoforms compared to mono‐articular vastii (∼62 vs . 54%) (Johnson et al., 1973 ) and these fibres preferentially atrophy into older age (Andersen, 2003 ; Hepple & Rice, 2016 ; Hunter et al., 2016 ). Further, sodium channel inactivation in rectus femoris in older females has been attributed to downregulation in Na + /K + ATPase (Lee et al., 2018 ) contributing to potential denervation (Clausen, 2003 ), although at a single muscle fibre level in vastus lateralis, α1 and α2 Na + /K + ATPase isoforms are not different between older and younger mixed‐sex cohorts (Wyckelsma et al., 2016 ). Although we could not reliably segment different quadriceps lean muscle CSAs, Mizuno et al. ( 2021 ) manually segmented quadriceps muscles (inclusive of lean and intermuscular fat mass) from computed tomography and reported comparable inter‐muscular reductions in whole CSA beyond 40 years of age in healthy females, whereas Maden‐Wilkinson et al. ( 2013 ) showed a greater degree of atrophy in rectus femoris compared to vastii muscles in older females. Thus, reduced M‐wave amplitude for rectus femoris may be explained by a contribution of both reductions in lean muscle CSA and denervation. Indeed, denervation occurs prior to muscle atrophy due to axonal sprouting and motor unit remodelling (Hunter et al., 2016 ). Rectus femoris activation during aerobic locomotor tasks (e.g. cycling, walking or jogging) of moderate to high intensity are much lower compared to vastii, with recruitment levels >90% MVC reached only during maximal intensity (O'Bryan et al., 2018 ), suggesting maximal intensity exercise may be necessary to mitigate rectus femoris neuromuscular degeneration in older age. Rectus femoris plays a significant role in torque generation and transfer across the hip and knee joints during locomotor functional tasks (Pandy & Andriacchi, 2010 ), illustrating its importance in the functional capacity of older adults. Although more investigation is warranted to dissociate rectus femoris denervation/atrophy from an increase in muscle–electrode distance arising from age‐related increases in subcutaneous and intramuscular fat mass (Akima et al., 2015 ; Mizuno et al., 2021 ; Nordander et al., 2003 ), we observed no decrease in vastii M‐wave amplitudes despite others reporting greater age‐related increases in vastus lateralis adipose tissue area compared to rectus femoris (Akima et al., 2015 ). Potential confounding effects of menstrual cycle phase on rectus femoris M MAX also warrant further investigation to validate our interpretation. PC 1 explained 25% of the variance in the data and was positively correlated with MVC. This dimension included moderate to strong positive correlations with oestradiol and progesterone; moderate correlations with bone density, MVPA, quadriceps lean CSA, protein intake and free oestradiol index (i.e. bioactive and not bound to sex hormone‐binding globulin); and a moderate to strong negative correlation with age. Aside from the well‐described negative consequences of advancing age, this finding illustrates the interplay between lifestyle factors (physical activity and protein intake), lean mass, bone density and primary female sex hormones on the capacity for maximal voluntary isometric torque production. Skeletal muscle and bone are tightly interconnected, with internal forces stimulating dose‐dependent changes in bone formation (Rubin & Lanyon, 1984 ) and various skeletal muscle myokines and growth factors (e.g. insulin‐like growth factor‐1, interleukin‐6), which influence muscle protein synthesis and bone turnover rate (Gomarasca et al., 2020 ). Further, both musculoskeletal structure and function are sensitive to oestrogens (Alexander et al., 2022 ; Almeida et al., 2017 ), physical activity (Fuchs et al., 2024 ; O'Bryan et al., 2022 ) and dietary protein (Deutz et al., 2014 ). At physiological concentrations, we reported that progesterone may promote muscle protein synthesis and oestradiol may improve muscle contractile function (Alexander et al., 2022 ; Critchlow et al., 2023 ). Further, others have shown oestradiol and progesterone replacement therapies may be beneficial to voluntary muscle strength (Greising et al., 2009 ) but not muscle mass (Javed et al., 2019 ), suggesting potential positive effects of these sex hormones on skeletal muscle quality and/or motor unit recruitment. Approximately 16% of the variance in the data was explained by PC 3, which was negatively correlated with PT 100 and PT 10 . PC 3 included moderate negative correlations with MVPA, quadriceps lean CSA and testosterone; and moderate positive correlations with total and free oestradiol index and quadriceps intramuscular fat. These results suggest that individuals with lower oestradiol but higher androgen levels after menopause may have a mitigated loss of muscle function arising from physiological mechanisms occurring at or distal to the neuromuscular junction. Thus, this finding may suggest a potential compensatory mechanism whereby low oestradiol may be counteracted by increasing androgens to maintain intrinsic muscle function. Indeed, declining ovarian function during menopause decreases the circulating concentrations of multiple hormones, but especially oestradiol, increasing the testosterone to oestradiol ratio (Burger et al., 2008 ; Pöllänen et al., 2011 ). The relative importance of free oestradiol and androgens is further highlighted by moderate and strong correlations to PC 2 (which explained 18% of the variance), although this was not correlated to any of the measured neuromuscular variables. Thus, bioavailable forms of these hormones may influence underlying neuromuscular mechanisms not captured by the present study, including supraspinal or spinal circuits (Piasecki et al., 2024 ), or molecular processes within skeletal muscle fibres (Alexander et al., 2025 ). Although the association between reduced evoked torque and PC 3 included lower total testosterone, total testosterone has been shown to have minimal effect on lean mass or voluntary muscle strength in premenopausal (Alexander et al., 2021 , 2025 ) and postmenopausal females (Alexander et al., 2022 ). Overall, further investigation into the role of different forms of primary sex hormones in neuromuscular ageing across the female lifespan is warranted. Finally, PC 4, explaining 10% of the variance, was associated with lower isometric voluntary and evoked torques, and was explained by lower quadriceps CSA, quadriceps intramuscular fat, free oestradiol index and higher protein intake. These findings support the role of higher quadriceps lean CSA and free oestradiol index for improved overall isometric voluntary and evoked torque production but indicate a potentially conflicting influence of lower intramuscular fat and higher protein intake. However, in this instance it is likely that the lower absolute quadriceps intramuscular fat may be representing overall reduced whole quadriceps area, and higher protein intake may be associated with increased fat intake as part of a generalized diet and potentially decreasing quadriceps force capacity (Arias‐Fernández et al., 2020 ). The contribution of alterations in female sex hormone concentrations to the accelerated decline in neuromuscular function from the fourth decade is strengthened by comparing our data to age‐related trajectories previously reported for males. For example, females lose more quadriceps isometric knee extension strength and dynamic power between 40 and 60 years of age compared to males (Roberts et al., 2018 ), with the magnitude of the sex‐related difference greater for dynamic power (Haynes et al., 2020 ). At a structural level, this may coincide with decreases in quadriceps lean CSA, muscle quality and type IIx myofibre distributions being more prominent in females than males (Mizuno et al., 2021 ; Roberts et al., 2018 ). For males, age‐related neuromuscular dysfunction tends to accelerate during the fifth to sixth decade (Haynes et al., 2020 ; Lindle et al., 1997 ; Mizuno et al., 2021 ), with some studies reporting improved physical performance associated with elevated levels of bioavailable testosterone, dihydrotestosterone (DHT) and dehydroepiandrosterone (DHEA) (Hsu et al., 2016 ; O'Donnell et al., 2006 ). However, potential sex differences and the extent to which primary female and male sex hormones contribute to neuromuscular decline across the adult lifespan remain to be investigated. Previous evidence suggests that some measures of neuromuscular function may fluctuate with hormonal changes across the menstrual cycle. Downstream at the spinal motoneurons, presynaptic inhibition of Ia sensory afferents (Hoffman et al., 2018 ) and firing rate of vastus lateralis motor units (Piasecki et al., 2023 ) are reduced during the ovulation to mid‐luteal phase of the menstrual cycle. Premenopausal females also display higher persistent inward currents in several lower limb muscles compared to young males, suggesting a potential role of sex hormones in neuromodulation of spinal motoneurons (Jenz et al., 2023 ). While some have reported potential menstrual cycle effects on corticospinal and intracortical circuits (Ansdell et al., 2019 ), others have shown limited effects on skeletal muscle contractile function (de Jonge et al., 2001 ). Despite the potential confounding effects of these changes on motor unit recruitment and firing rate, most studies and reviews report no or negligible effects of the menstrual cycle or contraceptive use on acute isometric or dynamic maximal strength in premenopausal females (McNulty et al., 2020 ; Piasecki et al., 2023 ). Thus, although the phase and contraceptive effects reported herein are between participants, we suspect no or trivial effects of menstrual cycle phase or contraceptive use occurred within participants for the reported neuromuscular variables. Although small alterations in motor unit firing rate may have large consequences on the rate of force development during upper and lower limb muscle ballistic voluntary contractions in males (del Vecchio et al., 2019 ; Škarabot et al., 2024 ), more research adopting similar high‐density surface electromyography methodologies is required in females. This study is the first to evaluate female neuromuscular function evenly across each decade from 18 to 80 years of age and to attempt to associate any observed decline with inter‐individual differences in sex hormone concentrations. The onset of neuromuscular degeneration started during the fourth decade and menopause onset, with intrinsic peripheral muscle function reducing before any observed functional decline. Neuromuscular degeneration in ageing females may be more pertinent for rectus femoris than for other quadriceps muscles, suggesting potential vulnerability to ageing and functional impairment. Although some of the variance in the postmenopausal data was explained by differences in quadriceps tissue composition and lifestyle factors, all principal components involved significant correlations with total or free sex hormone concentrations (oestradiol, progesterone and/or testosterone), illustrating the potential importance of these primary sex hormones in neuromuscular function in ageing females. Thus, future research should explore how interventions aimed at addressing dramatic and sudden decreases in sex hormone concentrations around menopause onset may help mitigate age‐related neuromuscular degeneration across the female lifespan.

Neuromuscular function describes the integration and translation of multiple synaptic potentials generated and received by upper and lower motoneurons into force or torques produced by the musculotendinous unit, thereby encapsulating mechanisms spanning the entire motor pathway. Ageing has detrimental effects on neuromuscular function characterized by a decrease in maximal muscle strength or power, which may be associated with reduced motor unit recruitment or loss in skeletal muscle mass and quality (Hunter et al., 2016 ; O'Bryan & Hiam, 2022 ; Tieland et al., 2018 ). For example, isometric muscle strength declines by ∼1–5% per year between ∼60 and 80 years of age (Delmonico et al., 2009 ; Hughes et al., 2001 ; Kim et al., 2018 ) with older males and females greater than 60 years of age being ∼20–40% weaker than their younger counterparts (Lindle et al., 1997 ; Roberts et al., 2018 ; Wu et al., 2016 ). The decline in strength and power occurs earlier in the lifespan for females when compared to males (Haynes et al., 2020 ; Roberts et al., 2018 ) and is accelerated at a faster rate beyond 80 years of age (Kim et al., 2018 ). Moreover, females exhibit a higher frailty index (Gordon et al., 2017 ) leading to a larger percentage of older females accessing aged‐care facilities and requiring assistance with activities of daily living (Austad, 2006 ). Despite the evidence that females display different trajectories in the age‐related decline in neuromuscular function compared to males, the mechanisms responsible for this dimorphic response remain unclear largely due to the inherent underrepresentation of females in physiological research (Garcia‐Sifuentes & Maney, 2021 ). Distinct transient fluctuations in the synthesis and concentration of primary ovarian female sex‐hormones oestradiol and progesterone, but also testosterone, across the different phases of the female lifespan including premenopausal (∼18–45 years), perimenopausal (∼40–55 years) and postmenopausal (∼50 years onward) periods is one primary potential cause (Alexander et al., 2022 ; Critchlow et al., 2023 ; Hansen & Kjaer, 2014 ; Piasecki et al., 2024 ). Indeed, sex hormone‐sensitive nucleic or membrane‐bound receptors are found throughout both the central nervous (Barth et al., 2015 ) and musculoskeletal systems (Wiik et al., 2009 ). Animal models demonstrate a clear effect of steroidal sex hormones (or their precursors) on neuromuscular function (Barth et al., 2015 ; Callachan et al., 1987 ; Collins et al., 2019 ; Moran et al., 2007 ; Piasecki et al., 2024 ; Schultz et al., 2009 ; Smith et al., 2002 ). In ovariectomized female rats administration of progesterone potentiates GABA inhibition and suppresses glutamate‐induced excitation (Smith et al., 1987 ). Conversely, oestradiol (the predominant form of oestrogen) suppresses GABAergic neurons and attenuates the release of GABA in pyramidal cells (Schultz et al., 2009 ), enhances glutamatergic transmission (Barth et al., 2015 ), and possesses potential excitatory effects modulated by dopaminergic (Rey et al., 2014 ) and serotonergic pathways (Lu et al., 1999 ). Beyond the neuromuscular junction, ovariectomized mice with oestradiol deficiency demonstrate reduced skeletal muscle contractility (Moran et al., 2007 ), cross‐sectional area (CSA) (Moran et al., 2007 ) and satellite cell number in fast‐twitch skeletal muscles (Collins et al., 2019 ) that can be restored with oestradiol treatment. Although cause and effect relationships in humans are more difficult to define, female ageing studies provide an observational model to gain an understanding of the role of primary sex hormones on neuromuscular function. Indeed, during perimenopause and postmenopause, oestradiol reduces to one‐third and one‐fifth of that observed during the reproductive years, respectively, whereas progesterone reduces to two‐thirds the level of younger females during perimenopause and is essentially absent in postmenopause (Burger et al., 2008 ; Landgren et al., 2004 ). However, the potential role female sex hormones play in the observed decline in neuromuscular function in older females has not been investigated in a large cohort of participants. Quadriceps strength and power is associated with functional capability throughout the lifespan (Fuchs et al., 2023 ; Martien et al., 2015 ), leading several studies to investigate the effects of ageing and the menopausal transition on quadriceps neuromuscular function in females. Compared to young females, postmenopausal females less than 65 years of age generate lower (Pöllänen et al., 2015 , 2011 ) or comparable (Laakkonen et al., 2017 ) absolute or specific (per cross‐sectional area) isometric torque during quadriceps maximal voluntary contraction (MVC), with similar conflicting results reported for quadriceps muscle CSA (Ahtiainen et al., 2012 ; Collins et al., 2019 ; Juppi et al., 2020 ; Park et al., 2019 ; Pesonen et al., 2021 ; Pöllänen et al., 2015 ). However, in postmenopausal females beyond the age of 65 years, lower quadriceps MVC and CSA compared to younger females is more consistently reported (Häkkinen et al., 1996 ; Varesco et al., 2022 ; Wrucke et al., 2024 ; Wu et al., 2016 ; Yacyshyn & McNeil, 2020 ) and may be accompanied by reduced neural drive and voluntary activation (Mau‐Moeller et al., 2013 ; Rozand et al., 2020 ; Solianik et al., 2017 ; Wu et al., 2016 ), rate of torque development (Häkkinen et al., 1996 ; Wrucke et al., 2024 ; Yacyshyn & McNeil, 2020 ) and potentiated evoked torques (Solianik et al., 2017 ; Varesco et al., 2022 ; Wrucke et al., 2024 ; Yacyshyn & McNeil, 2020 ). However, many comparison studies between younger and older female cohorts are limited in their capacity to define the trajectory of neuromuscular changes across the lifespan and to investigate the critical perimenopausal period, which is marked by severe, unpredictable and fluctuating hormonal patterns (Burger et al., 2007 ; Landgren et al., 2004 ). Of the limited studies describing neuromuscular changes within the perimenopausal phase, subtle reductions in quadriceps isometric MVC and evoked twitch torque have been suggested (Pesonen et al., 2021 ). Larger cohort studies and scoping reviews of a wider age range of females covering each decade of the lifespan (20–80 years of age) suggest an accelerated decline in quadriceps MVC occurring around perimenopause (Haynes et al., 2020 ; Lindle et al., 1997 ) that may (Hughes et al., 2001 ; Mizuno et al., 2021 ) or may not (Delmonico et al., 2009 ; Rolland et al., 2007 ) be associated with reduced skeletal muscle mass. Thus, functional decline occurring during the perimenopause period may also be explained by several neuromuscular factors including alterations in motor unit recruitment (Piasecki et al., 2024 ), sarcolemmal excitability (Lee et al., 2018 ) and/or excitation–contraction coupling including Ca 2+ ‐mediated cross‐bridge formation (Mazara et al., 2021 ). Currently, age‐related trajectories of neuromuscular dysfunction across the female lifespan and including premenopausal, perimenopausal and postmenopausal periods remain to be thoroughly investigated. The aims of this study were to quantify age‐related changes in quadriceps neuromuscular function across the female lifespan and to investigate how inter‐individual differences in body composition, lifestyle factors (physical activity level and protein intake) and primary sex hormones (oestradiol, progesterone and testosterone) contribute to any observed postmenopausal decline. To address this aim, voluntary and evoked isometric quadriceps torque and surface electromyography outcomes, one‐repetition maximum leg press, body composition including isolated quadriceps lean and intramuscular fat CSA, and female sex hormone concentrations were measured in 88 apparently healthy females 18–80 years of age.

None.

Peer Review History