Leukocyte telomere length is associated with MRI‐thigh fat‐free muscle volume: data from 16 356 UK Biobank adults

Abstract Background Telomere attrition may share common biological mechanisms with bone and muscle loss with aging. Here, we investigated the association between these hallmarks of aging using data from UK Biobank, a large observational study. Methods Leukocyte telomere length (LTL as T/S ratio) was measured using a multiplex qPCR assay at baseline (2006–2010). Bone mineral density (whole body and regional; via dual‐energy X‐ray absorptiometry), trabecular bone score (via lumbar‐spine dual‐energy X‐ray absorptiometry images), fat‐free muscle volume (thighs; via magnetic resonance imaging), and muscle fat infiltration (thighs; via magnetic resonance imaging) were measured during the imaging visit (2014–2018). Regression models were used to model LTL against a muscle or bone outcome, unadjusted and adjusted for covariates. Results A total of 16 356 adults (mean age: 62.8 ± 7.5 years, 50.5% women) were included. In the fully adjusted model, thigh fat‐free muscle volume was associated with LTL in the overall sample (adjusted standardized β (aβ) = 0.017, 95% CI 0.009 to 0.026, P < 0.001, per SD increase in LTL), with stronger associations in men (aβ = 0.022, 95% CI 0.010 to 0.034, P < 0.001) than in women (aβ = 0.013, 95% CI 0.000 to 0.025, P = 0.041) (sex‐LTL P = 0.028). The adjusted odds ratio (aOR) for low thigh fat‐free muscle volume (body mass index‐adjusted, sex‐specific bottom 20%) was 0.93 per SD increase in LTL (95% CI 0.89 to 0.96, P < 0.001) in the overall sample, with stronger associations in men (aOR = 0.92, 95% CI 0.87 to 0.99, P = 0.008) than women (aOR = 0.93, 95% CI 0.88 to 0.98, P = 0.009), although the sex difference was not statistically significant in this model (sex‐LTL P = 0.37). LTL was not associated with bone mineral density, trabecular bone score, or muscle fat infiltration in the overall or subgroup analyses (P > 0.05). Conclusions LTL was consistently associated with thigh fat‐free muscle volume in men and women. Future research should investigate moderating effects of lifestyle factors (e.g., physical activity, nutrition, or chronic diseases) in the association between LTL and muscle volume.


Introduction
Global aging represents a serious challenge for healthcare in the 21st century, with all bodily systems facing longer lifespans.The musculoskeletal system is crucial for maintaining posture and mobility, and profound losses of bone and muscle mass during aging are associated with poor health outcomes (i.e., hip fractures). 1,2Such outcomes cause huge socioeconomic strain on societies worldwide.
At the heart of all bodily systems lies the cellular and molecular machinery needed to maintain the structure and function of that tissue.Identifying biological markers of aging which reflect cellular senescence has been put forward as a priority to extend health span. 3Telomeres, which are found at the end of chromosomes in the cell's nucleus and contain specific DNA sequences, have seen an increase in research pertaining to aging and aging-related diseases. 4Telomeres function to (i) organize the 23 chromosomes pairs in sequence, (ii) maintain genome stability by acting as a protective scaffold for chromosomes, and (iii) to ensure DNA is not lost/damaged during DNA replication. 5uring each cell division, telomeres undergo natural shortening until a point where the cell can no longer replicate and apoptosis occurs leading to tissue loss. 4,5Another factor known to accelerate telomere loss is oxidative stress caused by chronically elevated levels of reactive oxygen species that damage lipids, proteins and DNA. 4 This can lead to defective bone remodelling and proteolysis in skeletal muscle. 1 Lifestyle factors (i.e., physical inactivity, poor nutrition, and excess alcohol/smoking) and chronic diseases (i.e., cancers) are key instigators for an increase in oxidative stress in the body. 1 Telomeres are active in mesenchymal stem cell populations including myocytes, neurons, fibroblasts and adipocytes.These cell types play a biological role in tissue replication and formation (e.g., myocytes in myogenesis [muscle formation], fibroblasts in osteogenesis [bone formation], and adipocytes in adipogenesis [fat formation]). 1 Although telomere length is shorter in leukocytes and longer in skeletal muscle and fat tissues, age-related telomere shortening is moderately correlated between these cells/tissues. 6A recent meta-analysis of healthy adults supports this notion showing that telomere length measured in one tissue correlates with another. 7It is for this reason that leukocyte telomere length (measured in peripheral blood) is suggested as a non-invasive whole body marker for biological aging. 8Thus, leukocyte telomere length, as a marker of cellular senescence, may be linked with low bone and muscle mass or muscle fat infiltration in older adults.However, there is a scarcity of large-cohort studies examining this thesis as recently highlighted by reviews on the role of telomere dysfunction in age-related diseases such as osteoporosis 9 or sarcopenia. 8n a previous study, we investigated associations between LTL and osteosarcopenia (bone fragility [osteopenia/osteopo-rosis by [WHO criteria] and muscle weakness [sarcopenia using SDOC/EWGSOP2]) in 20 400 adults in the UK Biobank. 10 However, we found no association between LTL and osteosarcopenia or its components including low bone mineral density (BMD), low grip strength, or low appendicular lean mass adjusted for height squared. 10Nonetheless, LTL was inversely associated with slow walking speed. 10We discussed possible explanations for these findings including the possible inaccuracy of the measures used to quantify lean (muscle) mass, which were based on dual-energy X-ray absorptiometry (DXA) scans.
In this subsequent analysis of UK Biobank participants, we seek to examine the link of LTL with bone and muscle quality using more accurate measures of muscle size [i.e., muscle volume from magnetic resonance imaging (MRI)], bone density (i.e., trabecular bone score), and muscle quality [i.e., muscle fat infiltration (MFI) from MRI] when compared with our previous analysis. 10Elucidating the role of telomeres in the mechanisms underlying bone and muscle loss has implications for future diagnostic and therapeutic approaches aiming to improve the health of older adults.

UK Biobank
Over 500 000 participants were recruited in the UK Biobank (UKB) from 2006 to 2010 with ages between 40 and 70 years. 11,12Participants visited one of 22 assessment centres near residence when they completed online tests or questionnaires and conducted a range of physical assessments (see here: https://biobank.ndph.ox.ac.uk/showcase/ label.cgi?id=100006).Samples of blood, urine, and saliva were also collected.In 2014, UKB started the world's largest multimodality imaging study, 13 aiming to reinvite 100 000 participants to undergo brain, cardiac and whole body MRI, DXA, and carotid ultrasound.Currently, about 50 000 participants have been recruited and part of their MRI and DXA data were used in this project.

Included samples
All the active UKB participants attending the first imaging visit were included (n = 48 994).Participants with any missing values in LTL (exposure), bone or muscle measures (outcomes), or covariates were excluded, leaving a total of 16 356 samples (Figure 1).
Figure 2 shows the study design.LTL was measured at recruitment (baseline).Bone and muscle measures were collected at first imaging visits that occurred 7.9 years (±1.5) after baseline on average.Baseline covariates included demographic, socioeconomic, lifestyle factors, and disease states.Age and body mass index (BMI) were collected at first imaging visits.Data were extracted using UKB field IDs as listed in Table S1.

Leukocyte telomere length
Relative LTL was measured from peripheral blood leukocytes using a multiplex qPCR assay, as the ratio of telomere repeat copy number (T) relative to that of a single copy gene (S). 14echnical laboratory parameters (e.g., enzyme, primer batch, PCR machine, pipetting robot, operator, temperature, humidity, time of day, and/or extraction method) that impacted this technique were adjusted for before the release of the data.For further information on the exact technical parameters adjusted for, and reducibility and validity of this technique. 14
MRI data to quantify body composition were from scans using a Siemens Aera 1.5 T scanner (Syngo MR D13) (Siemens, Erlangen, Germany). 15Image analysis was performed using AMRA Researcher (AMRA Medical AB, Linköping, Sweden).Total thigh fat-free muscle volume (FFMV) was calculated as the total volume of all voxels with fat fraction <50%  ('viable muscle tissue') in the left and right anterior and posterior thighs. 16,17Mean fat infiltration (MFI) was calculated as the mean fat fraction in the 'viable muscle tissue' (FFMV) of the right and left anterior thighs. 16,17Low FFMV was defined as FFMV with BMI-adjusted residuals in the lowest sex-specific 20%.Similarly, high MFI was defined as MFI with BMI-adjusted residuals in the highest sex-specific 20%.

Covariates
Age and BMI were collected at the first imaging visit.Time difference between the baseline visit and the first imaging visit was calculated.Other covariates were assessed at baseline via online questionnaires or linkages to electronic health records.Demographic information included self-reported sex (male or female), ethnicity (White, Black, South Asian, or Other), education (from none to college or university degree), and area-based Townsend deprivation index, where higher scores indicate higher levels of material deprivation.Lifestyle factors included smoking status (never, previous, or current), physical activity classifications (low, moderate, or high) based on an adapted short form of the International Physical Activity Questionnaire (IPAQ), 18 and alcohol intake frequency (never, special occasions only, one to three times a month, once or twice a week, three or four times a week, daily or almost daily).Disease states at baseline were determined using the UKB cancer registries data and first occurrence data, which integrated primary care data, hospital inpatient data, death register records, and self-reported medical condition codes based on ICD-10 codes: cancer excluding non-melanoma skin cancer and inflammatory diseases including coronary heart disease, type 2 diabetes, and chronic kidney disease.

Statistical methods
A descriptive analysis was conducted to summarize variables plus muscle and bone combination groups for all the included samples and by sex.Men and women were compared with respect to each variable using a Wilcoxon rank-sum test for continuous variables and a chi-square test for categorical variables.Prior to association analysis, LTL, continuous muscle and bone outcomes were z-transformed using the rank-based inverse normal transformation.LTL was associated with each continuous muscle or bone outcome using a linear regression model, where the non-linearity of LTL was tested using a penalized cubic function in the framework of generalized additive model (GAM), with the basis dimension k = 10.Logistic regression models were used for low FFMV (reference: normal FFMV) and high MFI (reference: normal MFI), and multinomial logistic regression models for categorical outcomes with more than two levels, that is, muscle and bone combination groups (reference group: normal femoral neck BMD and normal FFMV or normal MFI).Sex and age are associated with LTL. 14 We tested the interactions of LTL with sex and age group (≥60 or <60 years) on the outcomes adjusting for covariates.A significant interaction result (P < 0.05) was followed by a subgroup analysis by sex or age group.All the models above were adjusted for baseline covariates (age at first imaging visit, sex, ethnicity, education, Townsend deprivation index, BMI, physical activity via IPAQ activity group, smoking status, alcohol intake frequency, cancer, coronary heart disease, type 2 diabetes, chronic kidney disease, and the time difference between the baseline visit and the first imaging visit).Sex was excluded from sex-specific analyses.BMI was excluded for binary muscle outcomes and muscle and bone combination groups as BMI had been adjusted to determine sex-specific low FFMV and high MFI.P-values smaller than 5% were considered statistically significant.All the statistical tests are two-sided tests.The statistical analyses were performed in R version 4.2.2.Graphical illustration of the study design was created using BioRender (Science Suite Inc.) software.

Participant characteristics of the included samples
Table 1 shows participant characteristics overall and by sex.Among 16 356 adults, 49.5% were men and 50.5% were women.The majority were 60 years and older (66%), White (97%), and well educated (46% had a college or university degree).They attended first imaging visits at the mean age 62.79 years (±7.5).Data of LTL and most of the covariates were collected 7.9 years (±1.5) on average, prior to the first imaging visit.Participants had higher socioeconomic status and healthier lifestyles than the general population, with the mean BMI 26.5 (±4.3) kg/m 2 , and the mean Townsend deprivation index À2.08 (±2.6) versus the population average zero.Sixty-one per cent were never smokers.5% had never drunk and 22% drank daily or almost daily.The mean LTL (T/S ratio) was 0.84 (±0.1) after adjusting for the influence of technical parameters, higher in women (0.85 ± 0.13) than in men (0.82 ± 0.13) (P < 0.001).Men had higher mean BMD and L1-L4 trabecular bone score than women, for example, À0.63 ± 1.01 in men versus À0.69 ± 1.08 in women for femoral neck BMD T-score (P < 0.001).Compared with women, men also had higher mean FFMV (12.41 ± 1.73 g/ cm 2 in men vs. 8.28 ± 1.16 g/cm 2 in women, P < 0.001) and lower mean MFI (6.74 ± 1.68 g/cm 2 in men vs. 7.74 ± 1.81 g/cm 2 in women, P < 0.001).

Muscle and bone combination groups
In Table S2, the prevalence of osteoporosis at the first imaging visit was about 1% (n = 102) versus 15% (n = 2494) of osteopenia overall and similarly in men and in women.By definition, low FFMV and high MFI were controlled at 20% in both men and women.Due to the low number of osteoporotic cases, osteopenic and osteoporotic cases were not separated to make muscle and bone combination groups with binary FFMV or MFI.As a result, low FFMV or high MFI and osteopenic/osteoporotic groups accounted for 5% of the samples, overall and by sex (Table S2).

Associations between leukocyte telomere length and muscle or bone outcomes
The linearity for the relationship between LTL and a muscle or bone outcome was justified by the fitted GAM smoothing curves in men and in women: (1) residuals of z-transformed FFMV after regressing out the effect of BMI versus z-transformed LTL (Figure S1), (2) residuals of z-transformed MFI after regressing out the effect of BMI versus z-transformed LTL (Figure S1), (3) z-transformed bone outcomes versus z-transformed LTL (Figure S2).Additionally, we tested non-linearity for each outcome by comparing the model with a penalized cubic term of LTL and covariates and that with a linear term of LTL and covariates using an ANOVA F-test.None of the test results across the outcomes and samples were statistically significant (P > 0.05); therefore, the relationships between LTL and muscle or bone outcomes were modelled using a linear term of LTL in linear regression models.
There was no significant interaction between LTL and age group for any of the outcomes.The association with LTL significantly differed between men and women consistently across the outcomes and we highlighted significant sex-specific associations only.Longer LTL was significantly associated with higher FFMV (P < 0.001).The mean FFMV was increased by 0.017 SD (aβ = 0.017, 95% CI 0.009 to 0.026, Table 3) per SD increase in LTL after adjusting for covariates, more in men (aβ = 0.022, 95% CI 0.010 to 0.034, P < 0.001) than in women (aβ = 0.013, 95% CI 0.000 to 0.025, P = 0.041) (sex-LTL P = 0.028) (Table 2).In contrast, LTL was not significantly associated with any of the bone outcomes (femoral neck BMD, total body BMD, leg BMD, and L1-L4 TBS) or MFI overall, in men, and in women (Table 2).
In the model diagnostics analysis, the standardized residuals showed no significant deviation from the expected values under the assumption of a normal distribution (Figure S3).Additionally, the plot of standardized residual versus fitted value did not indicate any heteroscedasticity or discernible pattern (Figure S4).Although there were outliers with standardized residuals outside the range of À3 to 3, the results remained similar after removing the outliers (Table S3 vs. Table 2 -All).The variance inflation factors were low around 1 across predictors and models, suggesting no significant multicollinearity (Figure S5).Consistently, LTL was inversely associated with low FFMV.The adjusted odds ratio (aOR) for low FFMV was 0.93 per SD increase in LTL (P < 0.001), lower in men (aOR = 0.92, 95% CI 0.87 to 0.99, P = 0.008) and higher in women (aOR = 0.93, 95% CI 0.88 to 0.98, P = 0.009), but the sexdifference was not statistically significant (sex-LTL P = 0.37).LTL was not significantly associated with high MFI and osteopenic/osteoporotic, overall, in men, and in women.The aOR for high MFI was 1.00 (95% CI 0.96 to 1.04, P = 0.914) per SD increase in LTL.In regard to femoral neck BMD, the aOR of being osteopenic/osteoporotic versus normal was 1.01 (95% CI 0.98 to 1.05, P = 0.459) per SD increase in LTL (corresponding association results for men and women, respectively, in Table 3).In the model diagnostics analysis, deviance residuals fell within the range of À3 to 3, suggesting the absence of outliers (Figure S6).Furthermore, no discernible pattern emerged against the assumption of homoscedasticity (Figure S6).There was no evidence of a significant level of multimorbidity (Figure S7).
In the joint analysis of muscle and bone combination groups, longer LTL was significantly associated with low FFMV and normal femoral neck BMD.The adjusted relative risk (aRR) of low FFMV and normal femoral neck BMD versus normal FFMV and normal femoral neck BMD was 0.92 (95% CI 0.86 to 0.97, P = 0.003) overall and similarly in men (aRR = 0.93, 95% CI 0.86 to 1.01, P = 0.075) and in women (aRR = 0.90, 95% CI 0.83 to 0.98, P = 0.016) (sex-LTL P = 0.722) (Table 4).However, LTL was not significantly associated with other muscle and bone combination groups including low FFMV and osteopenic/osteoporotic.The finding was applied to all, men, and women (Table 4), which is understandable due to conflicting associations of LTL with low FFMV and osteopenic/osteoporotic. Longer LTL was not significantly associated with either high MFI or osteopenic/ osteoporotic (Table 3) and any of the combination groups based on MFI and femoral neck BMD, overall, in men and in women (Table 4).

Discussion
In this large observational study, we sought to examine the association between LTL and bone and muscle quality.Findings showed that LTL was consistently associated with thigh FFMV, with stronger associations in men than in women.However, LTL was not associated with BMD, trabecular bone score, or MFI.The biological hypothesis linking telomere attrition with low bone and muscle mass stems from the role of differentiated stem cells in osteogenesis and myogenesis. 1 Bone cells and myocytes are needed to maintain the structure and function of their respective tissues.Telomere shortening, as a marker of cellular senescence, will impede this inherent physiological process from occurring at least to some degree. 4,5xidative stress, which is induced by the release of free radicals (particularly reactive oxygen species), may exacerbate this process. 4,5Aging is linked to an increase free radicals in both bone and skeletal muscle and this is likely governed by lifestyle factors such as physical inactivity, poorer nutrition or chronic diseases. 1Healthy bone and muscle are also endocrine organs and release molecules (i.e., osteokines and myokines) which may support an anti-inflammatory environment imposed by aging and oxidative stress. 1 The above biological connections may explain our consistent associations between LTL and FFMV.The fact that LTL was not associated with bone measures (bone density or trabecular bone score) may reflect the fact that stem cells are more active in skeletal muscle versus bone, with the former being a soft tissue and the latter being a hard tissue.In other words, stem cells are more active in muscle turnover/repair than in bone remodelling and muscle compared with bone mass deteriorates at a faster and greater extent during aging. 1 When examining the associations between LTL and bone and muscle combination groups in the overall sample, it was surprising that significant differences were observed with the low FFMV and normal bone subgroup but not with the low FFMV and osteopenic/osteoporotic subgroup.However, given that both of these groups were trending in the same direction (i.e., inverse associations between LTL and the groups containing low FFMV), this difference may be explained by slight differences in sample sizes.In further support, the normal FFMV and osteopenic/osteoporotic subgroups were trending in the opposite direction to both groups containing low FFMV.As seen in the continuous analysis, low FFMV was consistently associated with LTL while bone outcomes were not.Thus, we believe low FFMV to be the driver here.
A very recent study including 5051 adults (45.9 ± 16.2 years old) from NHANES showed associations between LTL (T/S ratio) and appendicular lean mass (adjusted for height squared) measured by DXA. 19An earlier study in 2750 adults aged 60 years or older showed no relationship between LTL (T/S ratio) and bone density/osteoporosis in the Health ABC study. 20Previous studies, including ours, have either shown no association 10 or weak associations 21 between LTL and appendicular lean mass or bone density measured by DXA.The present study strengthens these findings and adds novelty by including a much greater sample size as well as inclusion of more accurate imaging biomarkers (fat-free muscle volume from MRI; muscle fat infiltration from MRI; and trabecular bone score from DXA images).Apart from our previous paper is minimal due to significant variation in the exposure.Third, telomere length in this study was measured from peripheral blood leukocytes, not only do peripheral blood leukocytes represent a highly heterogeneous population, but such measures may be less germane than tissue-derived analyses involving musculoskeletal tissues.LTL, however, is moderately correlated with telomere length measured in tissues such as muscle and fat. 6Future studies should consider these factors.

Conclusions
To conclude, in this large observational study, LTL was robustly associated with thigh FFMV, with stronger associations in men than in women.However, LTL was not associated with BMD, trabecular bone score, or MFI.Future research may investigate the moderating effects of lifestyle factors (such as physical activity, nutrition, or chronic diseases) in the association between telomere length and FFMV.

Figure 1
Figure 1 Flowchart of the final analytical sample.

Figure 2
Figure 2 Graphical representation of the study design.

Table 1
Population characteristics a Data from first imaging visits.

Table 2
Associations between LTL and continuous muscle or bone outcomes *Adjusted standardized β's and P-values from linear regression models adjusting for baseline covariates: age, sex, education, ethnicity, Townsend deprivation index, BMI, smoking status, alcohol intake frequency, IPAQ activity group, cancer, coronary heart disease, type 2 diabetes, chronic kidney disease, and the time difference between the baseline visit and the first imaging visit.