Positive Associations of Dietary Intake and Plasma Concentrations of Vitamin E with Skeletal Muscle Mass, Heel Bone Ultrasound Attenuation and Fracture Risk in the EPIC-Norfolk Cohort

The prevalence of sarcopenia, frailty and fractures is increasing. Prevention options are limited, but dietary factors including vitamin E have the potential to confer some protection. This study investigated cross-sectional associations between dietary and plasma concentrations of vitamin E with indices of skeletal muscle mass (SMM) (n = 14,179 and 4283, respectively) and bone density (n = 14,694 and 4457, respectively) and longitudinal fracture risk (n = 25,223 and 7291, respectively) in European Prospective Investigation Into Cancer and Nutrition (EPIC)-Norfolk participants, aged 39–79 years at baseline. Participants completed a health and lifestyle questionnaire, a 7-day diet diary (7dDD) and had anthropometric measurements taken. Fat-free mass (as a SMM proxy) was measured using bioimpedance and bone density was measured using calcaneal broadband ultrasound attenuation (BUA) and incident fractures over 18.5 years of follow-up. Associations between indices of SMM, BUA and fracture risk were investigated by quintiles of dietary vitamin E intake or plasma concentrations. Positive trends in SMM indices and BUA were apparent across dietary quintiles for both sexes, with interquintile differences of 0.88–1.91% (p < 0.001), and protective trends for total and hip fracture risk. Circulating plasma α- and γ-tocopherol results matched the overall dietary findings. Dietary vitamin E may be important for musculoskeletal health but further investigation is required to fully understand the relationships of plasma tocopherols.


Introduction
The prevalence of sarcopenia (low levels of muscle strength, muscle quantity/quality and physical function [1]), frailty and fractures is increasing in our aging society. In the UK, the number of older people is growing; in 2018, there were 1.6 million people aged 85 years and over; by mid 2043, this is projected to nearly double to 3 million [2]. In parallel, the number of sarcopenic patients will dramatically increase, adding to the already considerable resultant public health issues [3]. A recent prospective study by Sousa et al. [4] found that sarcopenia was independently associated with hospitalisation costs and with an estimated increase of 34% for patients aged ≥65 years. Approximately 520,000 fragility fractures occurred in the UK in 2017, with fracture-related costs of GBP 4.5 billion; these numbers are estimated to increase by 26.2% and 30.2%, respectively, by 2030 [5]. Losses in bone density and skeletal muscle mass and strength occur gradually from the age 30 years, with increasing rates of loss in those over the age of 60 years [6,7]. These conditions are currently difficult to treat and, therefore, maintaining skeletal muscle and bone health during aging is important.
It has long been established that a close physiological relationship exists between muscle and bone, which changes with aging, but more recently it has become apparent that this is not solely related to mechanical function [8,9]. Factors, such as myokines, that are secreted by muscle, including insulin-like growth factor 1 and fibroblast growth factor 2, have paracrine and endocrine effects which can affect bone repair and metabolism [10,11]. Additionally, bone secretes factors such as osteocalcin and connexin 43, that have direct effects on muscle [12,13]. Studies have shown that both sarcopenia and frailty are risk factors for fractures and falls [14][15][16][17][18].
Known determinants of muscle and bone aging [19] include modifiable lifestyle risk factors, such as cigarette smoking, low physical activity and poor diet [20]. Limited research exists, mainly in older adults, which suggests that vitamin E, a lipid-soluble, antioxidant vitamin, may also be protective with respect to muscle mass and frailty [21][22][23][24][25][26], as skeletal muscle is the organ with the highest consumption of oxygen in the body. Positive associations of vitamin E intake with bone mineral density (BMD) and fracture risk have also been reported in both men and women [27][28][29][30]. The major forms of vitamin E in food are the αand γ-tocopherols, and thus these are found in greater abundance than other tocopherols and tocotrienols in tissues. The predominant form of vitamin E in the body is α-tocopherol, which has tended to be the focus of research, although some research has been carried out on other tocopherols, in particular, γ-tocopherol [31,32]. A number of mechanisms have been suggested as to how vitamin E may slow down aging of skeletal muscle [33][34][35]. Reduction in oxidative stress is thought to be a mechanism by which vitamin E homologues protect bone but it has been reported that αand γ-tocopherol have opposing inflammatory functions and may uncouple bone turnover, such as by increasing bone resorption without affecting bone formation [36,37].
The current study therefore aimed to investigate the potential associations of reported dietary vitamin E intake (α-tocopherol equivalents), as well as plasma concentrations of both αand γ-tocopherol and the ratio of α:γ-tocopherol, with measures of skeletal muscle mass (SMM) and bone density status concurrently, in a large cohort general population cohort of middle-aged and elderly men and women. Additionally, both dietary and plasma concentrations of vitamin E were examined in relation to fracture risk during 18.5 years of follow-up.

EPIC-Norfolk Study Design
The Norfolk cohort of the European Prospective Investigation Into Cancer and Nutrition (EPIC-Norfolk) is part of the Europe-wide EPIC study, which involves over half a million people in ten countries [38] and was initially designed to investigate diet and the risk of developing cancer. Details of cohort recruitment, data collection and participant characteristics have been published previously [39]. In brief, participants aged between 39 and 79 years were recruited from General Practitioners' surgeries, based in the rural areas of Norfolk and market towns as well as the city of Norwich, from 1993 to 1997. Since virtually all the population of the UK are registered with a general practice through the National Health Service, general practice age sex registers act as a population sampling frame. This cohort at baseline was comparable to the UK national population with regard to many characteristics, including age, sex and anthropometry measurements, but it had a lower proportion of current smokers [40]. The study was approved by the Norfolk District Health Authority Ethics Committee (98CN01) and all participants gave written informed consent, according to the Declaration of Helsinki. Of the 30,445 men and women who consented to participate in the study (39% response rate), 25,639 attended a baseline health examination (1HE) between 1993 and 1997. Of these, 15,028 attended a second health examination (2HE) between 1998 and 2000.

Blood Analysis
At 1HE, a 42 mL sample of blood was collected in citrated and plain monovettes and stored in a refrigerator. The next day, blood samples were processed and stored at −196 • C as plasma and serum. Serum cholesterol was determined for the full cohort in a Norfolk laboratory using a RA 1000 Diagnostics (Bayer, Basingstoke, UK) instrument, and cohort concentrations ranged from 2.10 to 12.40 mmol/L. The vitamins αand γtocopherol were analysed on a cohort subset that consisted of a series of previous casecontrol studies, where cases were defined by incident cardiovascular disease or cancer and four matched, disease-free controls. Plasma concentrations were analysed at IARC, Lyon (France), using high-performance liquid chromatography for the vitamins. In our analyses, concentrations for α-tocopherol ranged from 0.71 the 106.54 µmol/L and from 0.03 to 9.85 µmol/L for γ-tocopherol; we excluded one participant for whom the ratio of α-tocopherol to γ-tocopherol was greater than 1000). Plasma αand γ-tocopherol concentrations were adjusted for cholesterol, as this is seen as a more reliable marker for vitamin E nutritional status [50] since tocopherols are transported via circulation through lipoproteins; adjusted concentrations are presented in µmol/mmol, calculated by dividing the plasma tocopherol concentrations (µmol/L) by total cholesterol (mmol/L).

Measurement of Confounding Variables
Data collected via two self-administered health and lifestyle questionnaires (HLQ1 and HLQ2), before the 1HE and 2HE, respectively, were used to establish classification of a number of variables. Family history of osteoporosis was categorised as yes or no; menopausal status (women only (2HE)) was categorised as premenopausal, perimenopausal (<1 year), perimenopausal (1-5 years) or postmenopausal; hormone replacement therapy (HRT) status (women only (2HE)) was categorised as current, former or never users. The use of statins and steroids at 2HE were categorised as yes or no. Smoking status (derived from HLQ2) (never, former, current) was derived from yes and no responses to the following questions "Have you ever smoked as much as one cigarette a day for as long as a year?" and "Do you smoke cigarettes now?". Self-reported physical activity (derived from HLQ1) was assessed using both occupational and leisure activities and individuals were assigned to one of four categories: inactive, moderately inactive, moderately active and active [51,52]. Occupational social class at 1HE was defined according to the Registrar General's classification. Nonmanual occupations were represented by codes I, (professional) II, (managerial and technical), and IIIa (nonmanual skilled) occupations while manual occupations were represented by codes IIIb (manual skilled), IV (partly skilled) and V (unskilled) occupations [53].

Statistical Analysis
All analyses were stratified by sex as significant differences in body composition, SMM and age-related changes in bone existing [44] between men and women. p < 0.05 was considered to be statistically significant in individual analyses. To minimise missing data exclusions, some missing values were recoded as follows: missing menopausal status data (2.8%) as premenopausal if age <50 years and never-user of HRT, or as postmenopausal if age >55 years or a current or former HRT user. Participants missing data for other variables in the multivariable model were excluded. Participants were excluded from analyses if they had missing or extreme BIA impedance values (<300 or >1000 ohms [54]), FFM < 25, or for participants with extremes of BMI (<14 or ≥36 kg/m 2 ), since bioelectrical impedance measures are considered unreliable at these levels [55].

Cross-sectional Analyses
Cross-sectional analyses were carried out using data from the 2HE, using dietary or plasma data from the 1HE; 14,179 participants had complete data for diet and muscle analyses, and 4283 had complete data for plasma and muscle analyses; the figures for BUA analyses are 14,694 and 4457, respectively ( Figure 1). Multivariable adjusted regression with ANCOVA was used to investigate differences in indices of SMM and calcaneal BUA across sex-specific dietary intake quintiles of vitamin E intake (mg α-tocopherol equivalents). Trend testing was achieved by treating the median values for quintiles as a continuous variable [56]. Each model was adjusted for important physiological, lifestyle, and dietary factors, known to influence risk in this population. For SMM, these included age, smoking status, physical activity, social class, energy intake, percentage energy from protein, corticosteroid and statin use, menopausal and HRT status in women; for BUA, these included age, BMI, family history of osteoporosis, menopausal and HRT status in women, corticosteroid use, smoking status, physical activity, Ca intake, total energy intake, and Ca-and vitamin D-containing supplement use. The data were also analysed to take the amount of vitamin E from supplements into consideration, as excluding supplements may underestimate total nutrient intake [57]. In separate analyses, indices of SMM and calcaneal BUA were investigated across sex-specific plasma concentration quintiles of α-tocopherol, γ-tocopherol and the ratio of α:γ-tocopherol, with the covariates described above, but excluding dietary intake data.

Longitudinal Analyses
Longitudinal analyses used data from the 1HE together with incident hospital-recorded fractures for the participants (all hip, spine and wrist fracture cases up to 31 March 2018); the mean follow-up time was 18.5 years (467,077 total person years), and was calculated as the time between an individual's 1HE and this cut-off date, or death if earlier. Data for diet and fracture analyses were available for 25,223 participants; data for plasma and fracture analyses were available for 7291 participants (Figure 1). Prentice-weighted Cox regression was used to investigate associations between incidence of fractures and sexspecific quintiles of dietary vitamin E intake (mg α-tocopherol equivalents), or plasma concentrations, using the same adjustments as for the BUA models. Missing values were treated in the same way as in the BUA models. Total risk for hip, spine or wrist fracture was calculated as the risk for the first occurrence of one of these fractures; this does not consider multiple fractures, and therefore the sum of the specific-site fracture incidences does not sum to the total.

Characteristics of the Study Population
Selected characteristics are summarised in Table 1, stratified by dietary analysis group and sex. Mean dietary and supplement-derived intakes of α-tocopherol equivalents (mg/day) are shown for the different study groups. In the dietary model analyses, numbers of men and women are similar for the SMM and BUA measures and intakes of dietary and supplement α-tocopherol equivalents are also similar. Dietary intakes of α-tocopherol equivalents are slightly lower in the fracture dietary analyses groups, as is the percentage of participants taking vitamin E-containing supplements and the amount of α-tocopherol equivalents obtained from these supplements. No UK Reference Nutrient Intake value [58] has been defined for vitamin E, although safe intakes of α-tocopherol equivalents have been set at 4 mg for men and 3 mg for women. Of the 11,427 men in the fracture dietary analysis group, 1.7% had an intake <4 mg (n = 194); 1.0% of the 13,796 women had an intake <3 mg (n = 132). However, the European Food Safety Agency (EFSA) Panel on Dietetic Products, Nutrition and Allergies (NDA) set Adequate Intakes (AIs) of α-tocopherol for adults as 13 mg/day for men and 11 mg/d for women [59]; it was felt that Average Requirements (ARs) and Population Reference Intakes (PRIs) could not be set for α-tocopherol. Only 30.7% of the men and 25.8% of the women in the fracture cohort met these AIs. The fracture dietary analyses groups had a higher percentage of current smokers, manual workers and physically inactive participants. Table 2 presents selected characteristics, stratified by plasma analysis group and sex. Mean dietary and supplement-derived intakes of α-tocopherol equivalents (mg/day) are shown for the different study groups, in addition to plasma tocopherol concentrations, unadjusted and adjusted for cholesterol. In the plasma model analyses, numbers of men and women are similar for the SMM and BUA measures and concentrations of the plasma tocopherols and intakes of dietary and supplement α-tocopherol equivalents are also similar. Concentrations of the cholesterol-adjusted plasma tocopherols are also similar in the fracture plasma analysis groups, although the ratio is slightly lower in both men and women. A plasma tocopherol concentration of at least 11.6 µmol/L, or a minimum tocopherol:cholesterol ratio of 2.25 µmol/mmol is considered to be the lowest satisfactory value; the dietary requirement of vitamin E is that which is necessary to keep the ratio above this level [58], although in a recent publication by EFSA, it was considered that they were insufficient data on markers of α-tocopherol intake/status/function (e.g., plasma/serum α-tocopherol concentration, markers of oxidative damage) to calculate the requirement for α-tocopherol [59]. Of the participants in the SMM plasma analysis group, only 4 men and 8 women had values <11.6 µmol/L; one of these woman had a tocopherol:cholesterol ratio <2.25 µmol/mmol. Eight men and 13 women had tocopherol:cholesterol ratios <2.25 µmol/mmol. Once again, the larger fracture plasma analysis groups had a higher percentage of current smokers, manual workers and physically inactive participants.

Food Sources of α-and γ-Tocopherols
Good sources of vitamin E include plant oils-such as rapeseed, sunflower, soya, corn and olive oil-nuts, seeds and wheatgerm. The main sources of γ-tocopherol include oils [60] (especially soybean and corn oils, which are used extensively in processed foods), nuts and seeds [61] (especially walnuts, pecans and pistachios, as well as sesame, flax and pumpkin seeds), as well as spinach, carrots, avocado, dark green leafy vegetables and wheatgerm. Figure 2 shows the main food group sources for men and women. Generally, the main food groups contributing to vitamin E intake in men and women were similar, with butters, spreads and margarines being the main contributors. Foods in the grains and cerealbased products groups include both sweet and savoury biscuits, cakes, pies and quiches.

Correlations between Dietary Vitamin E Intake and Plasma Concentrations
A number of weak but significant correlations were found between the dietary intake of α-tocopherol equivalents and plasma concentrations of α-tocopherol. Dietary intake of α-tocopherol equivalents was significantly correlated with plasma concentration of α-tocopherol in the SMM cohort in men (r = 0.079, p < 0.001, n = 2232), but not in women (r = 0.038, p = 0.084, n = 2051). In the BUA cohort, significant correlations were found in both men (r = 0.083, p < 0.001, n = 2300) and women (r = 0.044, p < 0.05, n = 2143). In the fracture cohort, dietary intake of α-tocopherol equivalents was significantly correlated with plasma concentration of α-tocopherol in both men (r = 0.105, p < 0.001, n = 3707) and women (r = 0.08, p < 0.001, n = 3551).
When the plasma concentration of α-tocopherol was adjusted for total cholesterol, the correlations with dietary intake were found to be slightly stronger. Dietary intake of α-tocopherol equivalents was significantly correlated with plasma concentration of α-tocopherol in the SMM cohort in both men (r = 0.136, p < 0.001, n = 2232) and women (r = 0.1203, p < 0.001, n = 2051). In the BUA cohort, significant correlations were found in both men (r = 0.131, p < 0.001, n = 2300) and women (r = 0.125, p < 0.001, n = 2143). In the fracture cohort, dietary intake of α-tocopherol equivalents was significantly correlated with plasma concentration of α-tocopherol in both men (r = 0.1565, p < 0.001, n = 3707) and women (r = 0.153, p < 0.001, n = 3551).

Figure 2.
Percentage composition of food groups for vitamin E intake of men and women in the EPIC-Norfolk cohort. "Butters, spreads and margarines" include reduced fat types. "Grains and cereal-based products" include rice and ricebased dishes, pasta, sweet and savoury flans, pies and quiches, biscuits, cakes, breads and bread rolls. "Vegetables and vegetable-based dishes" include raw and cooked vegetables, vegetable dishes and mixed salads. "Breakfast cereals" include porridge and muesli. "Potatoes, including products and dishes" include potatoes, potato products, dishes and salads. "Fish, including shellfish, products and dishes" includes shellfish and fish-based dishes. "Cooking fats and oils" include hard margarines, animal fats, vegetable fats and ghee. "Fruit" includes fresh, cooked and canned fruit. "Potato and cereal-based savoury snacks" include crisps and other potato-based snacks and bread/pastry type snacks. "Meat, including products and dishes" include unprocessed white and red meats (and products and dishes), processed meat products and offal, including products and dishes. "Nuts and seeds" include nuts, seeds and nut butters. "Dairy" includes milk, cheese, yoghurts and dairy-based desserts. "Other" includes sugar, herbs and spices and dietetic products. "Soups" include homemade, canned and reconstituted dried soups. "Sauces" include sweet and savoury sauces, gravies, salad dressings, pickles, chutneys and stuffing. "Eggs" include eggs and sweet and savoury egg dishes. "Beverages" include alcoholic and nonalcoholic drinks, such as tea, coffee, water, soft drinks, fruit juice and squashes. "Legumes" include beans, pulses and lentils, dried, cooked and canned, and legume-based salads.

Associations between Dietary Vitamin E Intakes and Indices of SMM
Significant positive associations were found between sex-specific quintiles of dietary vitamin E and FFM and FFM BMI (p-trend < 0.001 in both men and women), after adjustments for covariates (Table 3), with significant interquintile differences (Q5 versus Q1) in FFM of +1.0% (p < 0.001) in both men, and women, and in FFM BMI of +1.7% (p < 0.001) in men and +1.9% (p < 0.001) in women. The addition of the amount of vitamin E derived from supplements to the fully adjusted models did not alter the associations.

Associations between Plasma Vitamin E Concentrations and Indices of SMM
In general, similar linear trends were found for both men and women, with those across quintiles of αand γ-tocopherol tending to be in the same direction, and the trend for the ratio of α-tocopherol: γ-tocopherol was in the opposite direction (Table 4). Linear trends in both men and women were most apparent across quintiles of plasma γ-tocopherol. In adjusted plasma model analyses, there was a significant positive trend across both αand γ-tocopherol quintiles for FFM in men (p < 0.001) and women (p < 0.01). However, for FFM, a significant negative trend was found across quintiles of the ratio of α-tocopherol:γtocopherol in both men and women (p < 0.001). Whereas, across quintiles of the ratio of α-tocopherol: γ-tocopherol, significant positive trends were found for FFM in both men (p < 0.001) and women (p < 0.01), and significant negative trends for FFM BMI in both men and women (p < 0.001). In the adjusted model for FFM, significant differences were found between quintile 1 and quintiles 4 and 5 of plasma γ-tocopherol in women (p < 0.01 and <0.05, respectively), whereas in men, significant differences were found between quintile 1 and quintiles 2 (p < 0.01), 4 (p < 0.05) and 5 (p < 0.01) of the ratio of α-tocopherol:γ-tocopherol. In the adjusted model for FFM BMI , significant differences were found between Q1 and Q3 of plasma α-tocopherol in women (p < 0.05). Regarding plasma γ-tocopherol, significant differences were found between Q1 and Q5 of plasma in men (p < 0.01) and between Q1 and Q4 (p < 0.05) and 5 (p < 0.01) in women. In women, significant differences were found between Q1 and Q4 (p < 0.01) and Q5 (p < 0.05) of the ratio of α-tocopherol:γ-tocopherol.

Associations between Dietary Vitamin E Intakes and Bone Density Status
Mean calcaneal BUA values, stratified by sex and quintiles of dietary vitamin E, are shown in Table 5, for unadjusted data and the fully adjusted model. Significant positive associations across quintiles of dietary vitamin E intake were evident in both men and women, after adjustments for covariates (p-trend < 0.001 in both men and women). In the fully adjusted model, a significant difference was identified in men, for Q3 versus Q1 (+1.8%; p < 0.05). Further adjustment for the amount of vitamin E derived from supplements did not modify the associations.

Associations between Plasma Vitamin E Concentrations and Bone Density Status
Analysis of mean calcaneal bone density measures, stratified by sex and quintiles of plasma vitamin E concentrations, is shown in Table 6, for both the unadjusted and fully adjusted models. In both men and women, mean BUA measures tended to significantly increase across quintiles of plasma αand γ-tocopherol and decrease across quintiles of the ratio of α-tocopherol:γ-tocopherol (p < 0.001). No significant differences were found between quintile 1 and any of the other quintiles in the adjusted models, in either men or women.

Associations between Dietary Vitamin E Intakes and Fracture Risk
In the fully adjusted dietary model analyses, significant positive associations were evident between quintiles of vitamin E intake and risk for total fractures and hip fractures in both men and women (p < 0.001), but negative associations were found for wrist fractures in women (p < 0.01) ( Table 7). In men, both total fracture risk and wrist fracture risk were significantly lower in Q2 versus Q1 (0.79; 95% CI 0.64, 0.98; p < 0.05 and 0.51; 95% CI 0.29, 0.90; p < 0.05, respectively). In women, hip fracture risk was significantly lower in Q2 versus Q1 (0.81; 95% CI 0.66, 0.99; p < 0.05). The addition of supplement-derived vitamin E intake to the fully adjusted models did not alter the associations.

Associations between Plasma Vitamin E Concentrations and Fracture Risk
In the fully adjusted plasma vitamin E analyses, significant linear trends were found for risk of total and hip fractures and quintiles of plasma α-, γ-tocopherol and the ratio of α-tocopherol:γ-tocopherol in men (p < 0.05) ( Table 8). The risk for both total and hip fracture decreased across quintiles of plasma α-tocopherol and the ratio of α-tocopherol:γtocopherol, but increased across quintiles of γ-tocopherol for hip fracture risk and tended to decrease for total fracture risk. In women, in the fully adjusted plasma vitamin E analyses, significant linear trends were found for risk of total and hip fractures and quintiles of plasma α-, γ-tocopherol and the ratio of α-tocopherol:γ-tocopherol (p < 0.05). The risk for total and hip fractures tended to decrease across quintiles of plasma α-tocopherol and the ratio of α-tocopherol:γ-tocopherol but increase across the quintiles of γ-tocopherol. In men, spine fracture risk was significantly higher in Q3 versus Q1 (2.00; 95% CI 1.00, 4.00; p < 0.05) in the fully adjusted ratio of α-tocopherol:γ-tocopherol model. Total fracture risk in women was significantly higher in Q4 versus Q1 (1.29; 95% CI 1.01, 1.65; p < 0.05) in the fully adjusted plasma γ-tocopherol model. However, hip fracture risk in women was significantly lower in Q3 versus Q1 (0.61; 95% CI 0.41, 0.89; p < 0.05) in the fully adjusted ratio of α-tocopherol:γ-tocopherol model. Wrist fracture risk in women was significantly lower in Q5 versus Q1 (0.56; 95% CI 0.32, 0.98; p < 0.05) in the fully adjusted plasma α-tocopherol model.
The associations of dietary intake and plasma concentrations of vitamin E with SMM, BUA and fracture risk are summarised in Table 9. Table 6. Multivariate adjusted calcaneal BUA for EPIC-Norfolk participants, stratified by sex and quintiles of plasma α-tocopherol, plasma γ-tocopherol and α-tocopherol:γ-tocopherol ratio, adjusted for by blood cholesterol measurement.

Discussion
To our knowledge, this is the first study using data from a large population cohort of middle-and older-aged men and women in the UK to assess the associations between both dietary vitamin E intake and plasma vitamin E concentrations and indices of SMM, bone density status and fracture risk. Our results show significant positive associations between both dietary vitamin E intake and plasma concentrations of both serum cholesterol-adjusted αand γ-tocopherol and FFM and BUA, and generally significant positive associations for fracture risk. The associations found with vitamin E for bone density status and fracture risk are independent of vitamin D and calcium intake, which are both known to be relevant for bone health. The results from this study indicate protection for musculoskeletal health with higher intakes and blood concentrations of vitamin E.
In the EPIC-Norfolk study, the mean daily dietary intakes of vitamin E in 11,535 men and 13,972 women, assessed using a 7dDD and expressed in milligrams of α-tocopherol equivalents, were 11.62 (SD 5.24) and 9.27 (SD 3.78) mg [48] which is very similar to the mean daily intakes in the cohorts in these analyses. These dietary intakes are in agreement with those from other European countries [62,63]. Although in relation to the AIs defined by EFSA [59], only 31% of men and 26% of women met the AIs.
Most studies conducted in the US reported higher average plasma γ-tocopherol levels as compared to studies conducted in Europe [32]. This is likely explained by the fact that γ-tocopherol is the major form (approximately 70%) of vitamin E in the US diet [31]. Therefore, this study focuses on European comparisons. The mean plasma αand γtocopherol concentrations in our study cohorts are similar to those found in a healthy Irish adult population [63] and in the UK National Diet and Nutrition Survey (NDNS) [64]. More than 99% of both men and women in our study cohorts had higher plasma tocopherol concentrations or tocopherol:cholesterol ratios than the minimum satisfactory values [58].
Our study found a number of weak but significant correlations between the dietary intake of α-tocopherol equivalents and plasma concentrations of α-tocopherol, which has also been observed in a small German study [65] (n = 92; r = 0.14) but no associations have been found in a number of other European studies [66][67][68]. However, Kardinaal et al. found a significant age-and sex-adjusted correlation (r = 0.24, p < 0.05) for α-tocopherol between intake and adipose tissue levels in a small study of healthy adults [66], whereas no correlation was found between the adipose tissue level of alpha-tocopherol and dietary intake by Andersen et al. [67].
Significant positive trends in FFM and FFM BMI were evident across increasing quintiles of dietary vitamin E intake for both sexes, after adjusting for important covariates. Similar linear trends were generally apparent in both men and women for plasma vitamin E concentrations, with those across quintiles of αand γ-tocopherol tending to be in the same direction (positive for FFM but negative for FFM BMI ), and the trend for the ratio of α-tocopherol:γ-tocopherol in the opposite direction (negative for FFM but positive for FFM BMI ). These seemingly contradictory findings are not surprising as the increase in the ratio across quintiles is due to decreasing plasma αbut increasing γ-tocopherol concentrations. Whereas across increasing quintiles of plasma α-tocopherol, γ-tocopherol also increased, and across increasing quintiles of γ-tocopherol, plasma concentrations of α-tocopherol tended to decrease slightly. To date, most studies that have investigated the potential role of vitamin E in muscle health have focused on muscle function and strength rather than muscle mass and found that higher dietary vitamin E intakes [22,26] and plasma vitamin E concentrations were associated with higher strength measures and physical performance tests or lower levels of frailty [21,23,24,[69][70][71]. Findings from our study support the importance of vitamin E to skeletal muscle health.
Significant trends were apparent for BUA across quintiles of dietary vitamin E intake and plasma concentrations in both men and women, with BUA tending to increase across quintiles of dietary vitamin E intake and plasma αand γ-tocopherol but decrease across quintiles of the ratio of α-tocopherol:γ-tocopherol (p < 0.001). Significant positive associations were evident for dietary vitamin E intake and risk for total and hip fractures in both men and women (p < 0.001), but a significant negative association was found for wrist fracture risk in women (p < 0.01), where the greatest number of fractures was found in Q1, and the lowest in Q5; it is possible that this association may be artefactual. In plasma vitamin E analyses, significant linear trends were found for total fracture risk in men (p < 0.05) with the risk of total fractures generally decreasing across the quintiles in all 3 models. Significant linear associations were also found for hip fracture risk in men (p < 0.01), with the risk of hip fracture decreasing across quintiles of plasma α-tocopherol and the ratio of α-tocopherol:γ-tocopherol but increasing across quintiles of γ-tocopherol. In women, significant linear trends were found for risk of total and hip fractures and plasma vitamin E (p < 0.05), with the risk of fractures generally decreasing across quintiles of plasma α-tocopherol and the ratio of α-tocopherol:γ-tocopherol but increasing across quintiles of γ-tocopherol.
In the Aberdeen Prospective Osteoporosis Screening Study (APOSS) cohort, no biologically meaningful changes in BMD or bone resorption and formation markers with dietary intakes or serum concentrations of tocopherols or the α/γ ratio were found in perimenopausal and postmenopausal women [72], although dietary vitamin E intake was negatively associated with femoral neck BMD in early postmenopausal women in Scotland [73]. A recent study of nutrient intake and BMD in postmenopausal women found that a high intake of vitamin E had a negative effect on BMD [74]. Low serum concentrations of αtocopherol have been associated with an increased risk of hip fracture in elderly men and women [75] and an increased osteoporosis risk in postmenopausal women [27]. Both low intakes and serum concentrations of α-tocopherol were associated with an increased rate of fracture in elderly Swedish men and women [28]. There are a number of plausible explanations for the heterogeneity of the conclusions of the aforementioned epidemiological studies-the use of different covariates in the multivariable models, inconsistent measurement validity of biomarkers and the application of various exclusion criteria regarding the study sample-although most concur with our findings. Whether or not these study findings have any biological significance is unclear.
A recent review on the beneficial and detrimental effects of oxidative stress on human health concluded that αand γ-tocopherol forms of vitamin E exert a differential set of biological effects, which cannot always be regarded as positive to human health [76]. Recent data have also suggested that plasma α-tocopherol concentrations are more dependent on mechanisms that control circulating lipids rather than those related to its absorption and initial incorporation into plasma [77]; α-tocopherol was found to remain in circulation longer in participants with higher serum lipids, but its absorption was not dependent on the plasma lipid status.
In contrast to a high affinity to α-tocopherol (100%), α-tocopherol transfer protein (α-TTP) has a much lower affinity towards other vitamin E forms; 50%, 10-30%, and 1% affinity to β-tocopherol, γ-tocopherol, and δ-tocopherol, respectively [78], and plays an important role in the maintenance of high concentrations of α-tocopherol in plasma and some tissues [79,80]. A reduction in plasma γ-tocopherol during enhanced intake of α-tocopherol, such as through supplemental intake, can be explained by the more rapid metabolism of γ-tocopherol occurring when α-tocopherol intake is increased [81]. Chylomicron-associated tissue uptake of vitamin E may contribute to the accumulation of non-α-tocopherol forms of vitamin E such as γ-tocopherol in human skin, adipose tissue, and muscle, where unexpectedly high concentrations of γ-tocopherol were observed, in contrast to its low levels in the plasma [82]. Many unique properties have been attributed to γ-tocopherol and its metabolites [83], which exhibit sometimes enhanced or different activities of α-tocopherol such as natriuretic, anti-inflammatory, antitumoural activities, as summarised in a recent review [84]. The ratio of α-tocopherol:γ-tocopherol is suggested as a correction method as it would respond to even a small increase in α-tocopherol from supplementation that may not be clearly evident in plasma α-tocopherol concentrations [85]. Findings from the analyses in this study have shown that adjustment for the amount of vitamin E from supplements did not affect the associations.
The interactions between these two tocopherols are complex within the body and it must also be remembered that the bioavailability of vitamin E is influenced by a number of factors, including other nutrients, genetics, absorption, transport and metabolism [86]. With regard to other nutrients and food intake, data from the NDNS found that α-tocopherol correlated directly with "healthy" nutrient choices (intrinsic sugars, dietary fibre, and vitamins) and inversely with "unhealthy" choices (extrinsic sugars and monounsaturated fats-i.e., avoidance of polyunsaturated fat), whilst γ-tocopherol and the γ-tocopherol:αtocopherol ratio related inversely with "healthy" choices, with the authors concluding that the γ-tocopherol:α-tocopherol ratio may reveal poor dietary choices, which may subsequently lead to health issues in later life [87].
A number of possible mechanisms have been suggested illustrating how vitamin E may slow down the aging of skeletal muscle: (1) by improving antioxidant capacity, thereby reducing oxidative stress and inflammation; (2) improving membrane repair and increasing survival of damaged skeletal muscle by reducing oxidized phospholipid formation; (3) improving mitochondrial efficiency; (4) decreasing glycogen usage in skeletal muscle, while increasing fat metabolism; (5) enhancing muscle regeneration capacity; (6) stabilize insulin structure and improve insulin sensitivity of skeletal muscle [35]. It is thought that reduction in oxidative stress may also be a plausible mechanism whereby vitamin E protects bone, although the reported opposing inflammatory functions of αand γ-tocopherol may result in an increase in bone resorption without affecting bone formation [36]. However, further research is needed to investigate the potential effects of other tocopherols and tocotrienols on sarcopenic and osteoporotic risk factors.
The strengths of our study include a large population size of middle-aged and elderly men and women, from whom we had measures of dietary and supplemental intake, obtained from 7dDDs, in addition to plasma concentrations of vitamin E (α-and γ-tocopherol), in order to study the potential associations of vitamin E with indices of SMM, BUA and fracture risk (over 18.5 years of follow-up). Limitations of our study include the observational and cross-sectional study design regarding SMM and BUA measurements, precluding us from inferring causation, and the use of self-reported measures for dietary intake and physical activity. However, the prospective nature of our study of fracture risk and long follow-up for end points of 18.5 years are advantages. In addition, the 7dDDs developed for use in the EPIC-Norfolk study have previously been validated and are expected to produce a more precise measure of dietary intake than 24 h diet recalls or food frequency questionnaires [48]. Plasma vitamin E concentrations were only available for a small subset of the cohort, which may have reduced the power of our analyses. Nevertheless, the availability of plasma concentrations, which are not subject to nonrandom biases that can affect questionnaire-based measurements, is a strength of our study, although these concentrations may be affected by various physiological effects. SMMs were calculated from weight, height and bioelectrical impedance measurements, and not from potentially more accurate and precise methods, such as DEXA, computer tomography or magnetic resonance imaging; however, this method has comparable acceptability in population studies [88]. The dietary and lifestyle data, including the consumption of corticosteroids, HRT and dietary supplements, used in the longitudinal analyses were collected at 1HE and we were unable to account for any changes in exposures which may have occurred over time and potentially affected the associations.

Conclusions
Our research has found significant positive associations between greater intakes of dietary vitamin E and SMM indices, bone density status and total and hip fracture risk in both middle-aged and elderly men and women, with the scale of effects ranging from 0.88% to 1.91% (p < 0.001). Associations found with circulating plasma αand γ-tocopherol generally agreed with the dietary data. These findings suggest that dietary vitamin E intake may play a role in musculoskeletal health and provides evidence of the benefits of higher vitamin E intakes, similar to the AIs of α-tocopherol for adults of 13 mg/day for men and 11 mg/d for women, as recommended by EFSA. These intakes can be achieved by eating a varied and balanced diet, including the consumption of foods rich in vitamin E, such as oily seeds and their derivatives, nuts and cereals rich in vitamin E, including fortified breakfast cereals. Where it is not possible to obtain adequate intakes through the diet, vitamin E supplements should be consumed, especially in those at sarcopenic or osteoporotic risk. Further investigation is required to understand the relationships with plasma concentrations and musculoskeletal health.