Longitudinal alterations in lower extremity quantitative MRI metrics under simulated microgravity and their associations with systemic biomarkers

Abstract

This study employed a before-and-after design and utilized proton density fat fraction (PDFF) magnetic resonance imaging (MRI) to investigate the effects of 7 days of -6° head-down tilt bed rest (HDBR) on the lower limb muscles of healthy men. We performed multiple MRI scans of the participants’ thighs and calves during the HDBR period and collected blood and urine samples at corresponding time points to analyse biomarkers. We observed a sustained and significant reduction in thigh muscle cross-sectional area (TMCSA) and medial head of the gastrocnemius muscle cross-sectional area (MHMCSA) during HDBR (p < 0.001). The thigh subcutaneous adipose tissue area (TSAT) and the calf subcutaneous adipose tissue area (CSAT) showed a downward trend at the onset of HDBR, followed by a gradual recovery. During HDBR, both the thigh and calf muscle contractile mass index (CMI) exhibited a significant and sustained downward trend. Phosphorus (P), direct renin (DR), and red blood cells (RBC) showed significant correlations with muscle/fat parameters (p < 0.05). Our findings indicate that short-term HDBR leads to lower limb muscle atrophy and subcutaneous fat redistribution, and that these changes are dynamically associated with stress responses, metabolic indicators, and biomarkers related to bone metabolism.

Introduction

Spaceflight induces multisystem physiological deconditioning, with the musculoskeletal system being highly susceptible^1,2. Unloading of weight-bearing muscles precipitates rapid and significant skeletal muscle atrophy, functional impairment, and compositional remodelling, posing a critical risk to astronaut health. Head-Down Bed Rest (HDBR) is a well-established terrestrial analogy for simulating microgravity-induced adaptations, particularly muscle atrophy and functional decline³. Lower limb muscles are key to maintaining postural stability and movement under gravitational loading. During microgravity exposure, lower limb muscles decay at an accelerated rate due to interrupted weight-bearing cycles and reduced activity⁴. The HDBR model is particularly relevant for investigating the rapid deconditioning of lower-limb muscles.

While HDBR has been widely used to study inactivity-induced adaptations, critical knowledge gaps persist. Specifically, the development of muscle fat infiltration secondary to gravitational unloading of postural muscles has received limited attention, despite its recognized role as a key indicator of muscle health⁵. Previous studies have reported that intramuscular lipid droplets in the -6° head-down bed rest study began to accumulate by day 6⁶. In another 3-day dry immersion model, the muscle fiber cross-sectional area of intermuscular adipose tissue(IMAT) adipocytes tended to increase in comparison with the pre-dry immersion period⁷. The fat content of lumbar spine muscles increases in simulated microgravity conditions⁸. Elucidating the initial mechanisms of muscle fat infiltration is essential for developing effective intervention strategies. Existing studies have mostly focused on models of long-term disuse, while fat infiltration during the acute phase (<7 days) remains uncharacterized. Understanding the onset of fatty infiltration is critical, as it may represent an early, potentially irreversible step in muscle degradation. Concurrently, emerging evidence indicates spaceflight-induced haematological alterations^9,10, yet their association with skeletal muscle remodelling remains poorly understood. The ground-based studies have shown that indicators such as serum 25(OH)D and phosphate correlate with muscle status^11,12. Elucidating such relationships in microgravity could reveal novel mechanistic insights to inform muscle preservation strategies.

Magnetic Resonance Imaging (MRI) can accurately quantify changes in the lower muscles with advanced muscle structure assessment capabilities. Proton Density Fat Fraction (PDFF) enables objective assessment of intramuscular fat infiltration¹³. This technique offers superior reproducibility in tissue characterization compared to conventional methods¹⁴. Using PDFF, our study fills two key knowledge gaps: to describe and quantify acute changes in lower limb components (intramuscular/intermuscular fat, subcutaneous fat, bone marrow fat) during short-term HDBR; and to establish correlations between these muscular adaptations and biomarkers to improve understanding of the mechanisms responsible for the early changes in the components of the lower limb in the microgravity environment.

Results

Participant Characteristics

In this study, 44 participants underwent HDBR experiments. In thigh measurements, 10 participants were excluded because they dropped out of the experiment (n = 4) or missed the thigh measurements (n = 6) in calf measurements, 12 participants were excluded because they dropped out of the experiment (n = 4) or missed the calf measurements (n = 8). The participant characteristics are provided in Table 1.

Table 1.

Demographic characteristics of subjects in thigh group and calf group at pre-bed rest(PRE)

	Thigh	Calf
Sex, N	34, male	32, male
Age, years, mean ± SD	28.03 ± 6.74	27.47 ± 5.79
Height, cm, mean ± SD	171.6 ± 5.46	171.41 ± 5.56
Weight, kg, mean ± SD	68.21 ± 7.81	67.23 ± 6.89
BMI, mean ± SD	23.17 ± 2.55	22.90 ± 2.34

Longitudinal Analysis

The intraclass correlation coefficient (ICC) for MRI image segmentation was excellent. For cross-sectional area (CSA) measurements across different anatomical regions, the intra-rater reliability ICC ranged from 0.833 to 0.982, indicating good to excellent consistency in repeated measurements; the inter-rater reliability ICC ranged from 0.906 to 0.993, demonstrating excellent agreement among different raters.

A linear mixed model (LMM) was used to assess the longitudinal trajectories of Z-score-standardised MRI measures of the thigh at five time points (PRE, BR3, BR7, R3, and R5). Thigh muscle parameters exhibited significant longitudinal changes. Both thigh muscle cross-sectional area (TMCSA) and thigh muscle contractile mass index (TCMI) showed a sustained and significant decline during the bedrest period (PRE vs. BR3, PRE vs. BR7; both P < 0.001). Throughout the recovery period, these muscle parameters remained significantly below baseline levels (P < 0.05). Furthermore, significant differences in TMCSA and TCMI were observed between BR3 and BR7 (P < 0.001), as well as between the final day of bed rest (BR7) and the recovery period time points (R3, R5) (P < 0.01) (Table S1 and Fig. 1).

Fig. 1. Longitudinal changes in standardized thigh MRI parameters during bed rest and recovery.

Box plots display the median (horizontal line inside the box) and the interquartile range (IQR, top and bottom of the box). Individual colored dots represent the Z-score standardized data points for each participant at specific time points. Statistical significance between different time points was determined using estimated marginal means (EMMs) derived from linear mixed-effects models (LMMs), with Bonferroni correction applied for multiple post-hoc comparisons. *P < 0.05, **P < 0.01, ***P < 0.001. PRE baseline, BR3 bed rest day 3, BR7 bed rest day 7, R3 recovery day 3, R5 recovery day 5, TMCSA thigh muscle cross-sectional area, TCMI thigh contractile mass index, TSAT thigh subcutaneous adipose tissue, TSAT-FF thigh subcutaneous adipose tissue fat fraction, TTSMA-FF thigh total skeletal muscle fat fraction, TBMAT-FF thigh bone marrow adipose tissue fat fraction.

With regard to adipose tissue parameters, the subcutaneous fat area of the thigh (TSAT) decreased at the start of bed rest and then gradually increased. The inflection point for changes in the subcutaneous fat area of the thigh occurred on the third day of bed rest, whilst the inflection point for changes in the muscle area of the thigh occurred on the seventh day of bed rest. In contrast, the fat fraction of thigh subcutaneous adipose tissue (TSAT-FF) did not exhibit any statistically significant longitudinal changes across all experimental time points.Thigh bone marrow adipose tissue fat fraction (TBMAT-FF) showed significant differences between PRE and R5, BR3 and R5, BR7 and R3, and BR7 and R5 (P < 0.05) (Table S1 and Fig. 1).

A sensitivity analysis of longitudinal changes was conducted using a robust linear mixed model (Robust LMM) to verify the stability of the aforementioned longitudinal differences and to correct for potential outliers. The sensitivity analysis confirmed that, for all six assessed lower-leg MRI metrics, the standard LMM and the robust LMM exhibited 100% consistency and robustness in their assessments of the main effect of time (the overall longitudinal trend). Although the overall trajectory was extremely robust, some minor inconsistencies were observed between the two models when conducting specific post-hoc pairwise comparisons between time points. Specifically, the difference in total thigh skeletal muscle fat fraction (TTSMA-FF), which includes intermuscular fat (IMAT) and intramuscular fat (IMF), between BR3 and BR7 (standard model P = 0.043) was no longer significant in the robust analysis (robust model P = 0.161). Similarly, the initial significant difference in TBMAT-FF between PRE and R3 (standard model P = 0.033) was not supported by the robust model (robust model P = 0.053).

LMM was used to assess the longitudinal trajectories of Z-score-standardised MRI measures of the calf at five time points (PRE, BR3, BR7, R3, and R5). Significant longitudinal changes were observed in calf muscle parameters. Both calf muscle cross-sectional area (CMCSA) and calf muscle contractile mass index (CCMI) showed a marked decline during the bedrest period (P < 0.001). Throughout the recovery period, these muscle parameters remained significantly below baseline levels (P < 0.001). For CMCSA, a significant sustained decline was observed between BR3 and BR7 (P < 0.01), whereas a significant rebound was observed between the final day of bed rest and the recovery period time points (P < 0.001). The medial head of the gastrocnemius muscle cross-sectional area (MHMCSA) and the medial head of the gastrocnemius muscle contractile mass index (MHCMI) exhibited similarly pronounced longitudinal changes compared with baseline at all time points (P < 0.001). In contrast, changes in the lateral head of the gastrocnemius muscle cross-sectional area (LHMCSA) and the lateral head of the gastrocnemius muscle contractile mass index (LHCMI) were relatively mild, showing a significant decline from baseline to BR7 (P < 0.05) (Table S1 and Fig. 2).

Fig. 2. Longitudinal changes in standardized calf MRI parameters during bed rest and recovery.

Box plots display the median (horizontal line inside the box) and the interquartile range (IQR, top and bottom of the box). Individual colored dots represent the Z-score standardized data points for each participant at specific time points. Statistical significance between different time points was determined using estimated marginal means (EMMs) derived from linear mixed-effects models (LMMs), with Bonferroni correction applied for multiple post-hoc comparisons. *P < 0.05, **P < 0.01, ***P < 0.001. PRE baseline, BR3 bed rest day 3, BR7 bed rest day 7, R3 recovery day 3, R5 recovery day 5, CMCSA calf muscle cross-sectional area, CCMI calf muscle contractile mass index, MHMCSA medial head of gastrocnemius muscle cross-sectional area, MHCMI medial head of gastrocnemius calf muscle contractile mass index, LHMCSA lateral head of gastrocnemius muscle cross-sectional area, LHCMI lateral head of gastrocnemius calf muscle contractile mass index.

With regard to calf adipose tissue parameters, the calf subcutaneous adipose area (CSAT) decreased at the start of bed rest and then gradually increased. The inflection points for changes in calf subcutaneous fat area occurred on day 3 of bed rest, whilst the inflection point for changes in calf muscle area occurred on day 7 of bed rest. The calf subcutaneous adipose tissue fat fraction (CSAT-FF) increased significantly from PRE to BR3 and BR7 (P < 0.01), followed by a significant difference between the bed rest period and the recovery period (P < 0.01). For the intramuscular fat fraction, the standard LMM found a significant difference only between BR3 and R3 in the medial head of the gastrocnemius intramuscular fat fraction (MHIMF-FF) (P < 0.05). No statistically significant longitudinal changes were observed in the lateral head intramuscular of the gastrocnemius fat fraction (LHIMF-FF), calf bone marrow fat fraction (CBMAT-FF), or calf total skeletal muscle fat fraction (CTSMA-FF) between any time points (Table S1 and Fig. 3).

Fig. 3. Longitudinal changes in standardized calf MRI parameters during bed rest and recovery.

Box plots display the median (horizontal line inside the box) and the interquartile range (IQR, top and bottom of the box). Individual colored dots represent the Z-score standardized data points for each participant at specific time points. Statistical significance between different time points was determined using estimated marginal means (EMMs) derived from linear mixed-effects models (LMMs), with Bonferroni correction applied for multiple post-hoc comparisons. *P < 0.05, **P < 0.01, ***P < 0.001. PRE baseline, BR3 bed rest day 3, BR7 bed rest day 7, R3 recovery day 3, R5 recovery day 5, CSAT calf subcutaneous adipose tissue, CSAT-FF calf subcutaneous adipose tissue fat fraction, MHIMF-FF medial head of gastrocnemius intramuscular fat fraction, LHIMF-FF lateral head of gastrocnemius intramuscular fat fraction, CBMAT-FF calf bone marrow adipose tissue fat fraction, CTSMA-FF calf total skeletal muscle fat fraction.

Sensitivity analyses confirmed that, across all 12 lower-leg MRI measures evaluated, the standard LMM and robust LMM exhibited 100% consistency and robustness in their assessments of the main effect of time (overall longitudinal trend). Although the overall trajectory was extremely robust, some minor inconsistencies between the two models were observed when conducting specific post-hoc pairwise comparisons between time points. For example, the significant difference in MHIMF-FF observed between BR3 and R3 in the standard LMM (standard model P = 0.039) was no longer significant in the robust analysis (robust model P = 0.573). Conversely, some associations that were not significant in the standard LMM reached statistical significance in the robust LMM following algorithmic downweighing of outliers. These previously masked significant differences include: MHIMF-FF between PRE and BR7 (robust model P = 0.012), CSAT-FF between BR7 and R5 (robust model P = 0.024), CMCSA between BR3 and R5 (robust model P < 0.001), and CCMI between BR3 and BR7 (robust model P = 0.004) and between BR3 and R5 (robust model P < 0.001).

Regional differences in the rate of tissue changes in the thigh and lower leg

The rates of change in tissue parameters for the thigh and lower leg, along with the results of intergroup comparisons, are shown in Table 2. Notably, there was a significant regional difference in the rate of change in muscle cross-sectional area (ΔMCSA) (P = 0.009); ΔMCSA in the lower leg (-7.13 ± 3.84%) was significantly lower than that in the thigh (-6.05 ± 2.30%), suggesting a more severe degree of muscle atrophy in the calf. There was no statistically significant difference in the rate of change in subcutaneous adipose tissue area (ΔSAT) between the thigh and the lower leg (P = 0.310); both showed a slight downward trend, with mean rates of change of -5.13 ± 8.77% (thigh) and -3.48 ± 4.60% (lower leg), respectively. There were no significant differences between the thighs and calves in the rates of change in muscle-to-fat ratio (ΔTSMA-FF), subcutaneous fat ratio (ΔSAT-FF), muscle mass index (ΔCMI), and bone marrow fat ratio (ΔBMAT-FF) (P > 0.05) (Table 2).

Table 2.

Comparison of changes in tissue parameters in the thigh and calf (n = 32)

Parameter	Thigh	Calf	P-value
ΔMCSA, %	-6.05 ± 2.3	-7.13 ± 3.84	0.009
ΔCMI, %	-6.55 ± 2.59	-7.19 ± 4.42	0.132
ΔSAT, %	-5.13 ± 8.77	-3.48 ± 4.6	0.310
ΔSAT-FF, %	0.46 ± 1.38	0.84 ± 1.43	0.316
ΔTSMA-FF, %	3.59 ± 15.54	1.27 ± 10.55	0.607
ΔBMAT-FF, %	0.87 ± 3.91	0.06 ± 1.09	0.241

Δ, percentage change from baseline to day 7 of bed rest.

FF fat fraction, MCSA muscle cross-sectional area, SAT subcutaneous adipose tissue, TSMA total skeletal muscle adiposity, which includes intermuscular fat and intramuscular fat, CMI contractile mass index, BMAT bone marrow adipose tissue.

ΔSAT, ΔSAT-FF, and ΔBMAT-FF were analyzed by paired-samples t-test; ΔMCSA, ΔTSMA-FF, and ΔCMI were analyzed by Wilcoxon signed-rank test.

Association between biomarkers and MRI parameters

In addition to primary analyses, we conducted exploratory analyses on the relationship between MRI parameter changes and biomarkers.

LMM was employed to investigate the associations between systemic biochemical markers and MRI parameters of the thigh. Significant and widespread associations were observed between systemic biomarkers and thigh morphological indices. With regard to muscle parameters, TMCSA and TCMI were significantly negatively correlated with serum phosphorus (P), direct renin (DR), red blood cells (RBC), haemoglobin (Hb), cortisol (CORT), and urinary free cortisol (UCORT) (P < 0.0). Conversely, TMCSA and TCMI exhibited strong positive correlations with calcitonin (CT) and 25-hydroxyvitamin D [25(OH)D] (P < 0.01). TSAT showed significant negative correlations with DR, RBC, Hb, and UCORT (all P < 0.001). TSAT-FF was positively correlated with DR, adrenocorticotropic hormone (ACTH), RBC, white blood cell count (WBC), and Hb (P < 0.05). Similarly, TBMAT-FF was strongly positively correlated with RBC, Hb, and UCORT (P < 0.05). Furthermore, TTSMA-FF was significantly negatively correlated with 25(OH)D (P < 0.05). Several systemic biomarkers, including serum calcium (Ca) and osteocalcin (OC), did not show any statistically significant longitudinal associations with the assessed femoral MRI parameters (all P > 0.05). Furthermore, apart from the specific association with 25(OH)D, TTSMA-FF was largely independent of the longitudinal fluctuations in most systemic biochemical markers (Fig. 4).

Fig. 4. Heatmap of longitudinal associations between systemic biochemical markers and thigh MRI parameters.

The color gradient represents the standardized regression coefficients (β) derived from linear mixed-effects models (LMMs), where both the biochemical markers and MRI metrics were Z-score standardized prior to analysis to allow for direct comparison of effect sizes. Red gradients indicate positive associations, whereas blue gradients indicate negative associations; the intensity of the color reflects the magnitude of the effect size. Statistical significance is denoted by asterisks: *P < 0.05, **P < 0.01, ***P < 0.001. TMCSA thigh muscle cross-sectional area, TSAT thigh subcutaneous adipose tissue, TCMI thigh contractile mass index, TTSMA-FF thigh total skeletal muscle adiposity fat fraction (which includes intermuscular and intramuscular fat), TSAT-FF thigh subcutaneous adipose tissue fat fraction, TBMAT-FF thigh bone marrow adipose tissue fat fraction, Ca calcium, P phosphorus, DR direct renin, ACTH adrenocorticotropic hormone, IPTH intact parathyroid hormone, CT calcitonin, OC osteocalcin, 25(OH)D 25-hydroxyvitamin D, RBC red blood cells, WBC white blood cells, Hb hemoglobin, CORT cortisol, UCORT urinary cortisol.

To ensure that the above associations were not influenced by extreme intra-individual variability, sensitivity analyses were conducted using robust LMM in this study. Most of the associations identified in the standard LMM remained highly robust. However, specific differences were also observed: two associations were marginally significant in the standard model. TSAT-FF and DR (standard model, P = 0.011) and TTSMA-FF and 25(OH)D (standard model, P = 0.032) became non-significant in the robust model. Conversely, the algorithmic downweighing in the robust model revealed two significant associations that were not observed in the standard LMM and had been masked by outliers: TMCSA was negatively correlated with intact parathyroid hormone (IPTH) (robust model P = 0.023), and TTSMA-FF was positively correlated with ACTH (robust model P = 0.038).

This study employed a linear mixed model (LMM) to comprehensively assess the associations between all measured systemic blood biochemical parameters and local MRI parameters of the calf. There is a widespread and highly significant association between systemic biomarkers and lower leg morphological indices. P, DR, RBC, Hb, CORT, and UCORT all showed strong and consistent negative correlations with CMCSA, MHMCSA, CCMI, MHCMI, and CSAT (P < 0.01). For LHMCSA and LHCMI, only P, DR, and CORT maintained a significant negative correlation. Furthermore, WBC showed a specific negative correlation only with MHMCSA and MHCMI (P < 0.05). Conversely, the study also identified strong longitudinal positive correlations: calcitonin (CT) was positively correlated with CMCSA (P < 0.05), whereas 25-hydroxyvitamin D [25(OH)D] showed spectific positive correlations with LHMCSA and LHCMI (P < 0.01) (Fig. 5).

Fig. 5. Heatmap of longitudinal associations between systemic biochemical markers and calf MRI parameters.

The color gradient represents the standardized regression coefficients (β) derived from linear mixed-effects models (LMMs), where both the biochemical markers and MRI parameters were Z-score standardized prior to analysis to allow for direct comparison of effect sizes. Red gradients indicate positive associations, whereas blue gradients indicate negative associations; the intensity of the color reflects the magnitude of the effect size. Statistical significance is denoted by asterisks: *P < 0.05, **P < 0.01, ***P < 0.001. CMCSA calf muscle cross-sectional area, MHMCSA medial head of gastrocnemius muscle cross-sectional area, LHMCSA lateral head of gastrocnemius muscle cross-sectional area, CSAT calf subcutaneous adipose tissue, MHCMI medial head of gastrocnemius muscle contractile mass index, LHCMI lateral head of gastrocnemius muscle contractile mass index, CCMI calf muscle contractile mass index, CTSMA-FF calf total skeletal muscle adiposity fat fraction (which includes intermuscular and intramuscular fat), MHIMF-FF medial head of gastrocnemius intramuscular fat fraction, LHIMF-FF lateral head of gastrocnemius intramuscular fat fraction, CSAT-FF calf subcutaneous adipose tissue fat fraction, CBMAT-FF calf bone marrow adipose tissue fat fraction, Ca calcium, P phosphorus, DR direct renin, ACTH adrenocorticotropic hormone, IPTH intact parathyroid hormone, CT calcitonin, OC osteocalcin, 25(OH)D 25-hydroxyvitamin D, RBC red blood cells, WBC white blood cells, Hb hemoglobin, CORT cortisol, UCORT urinary cortisol.

With regard to lipid infiltration and tissue composition, systemic markers exhibited different patterns of association. DR, RBC, Hb, and UCORT were positively correlated with both CSAT-FF and MHIMF-FF (P < 0.05). Furthermore, IPTH was positively correlated with both MHIMF-FF and CTSMA-FF (P < 0.05). CTSMA-FF was also positively correlated with CT (P < 0.05) and negatively correlated with osteocalcin (OC) (P < 0.01). LHIMF-FF was positively correlated with UCORT and negatively correlated with OC (both P < 0.05). It is worth noting that Ca and ACTH did not show any statistically significant association with any lower leg MRI parameters. Similarly, CBMAT-FF was completely independent of longitudinal fluctuations in all assessed systemic biochemical markers (P > 0.05) (Fig. 5).

Sensitivity analysis of associations with calf biomarkers. To ensure that the associations were not influenced by extreme intra-individual variation, sensitivity analyses were performed using robust LMMs. Whilst the standard LMM effectively captured the main trends, the algorithmic weighting in the robust models revealed several key differences. Four associations initially identified as significant in the standard model became non-significant in the robust analysis (false positives): MHIMF-FF with IPTH, CTSMA-FF with CT, and LHIMF-FF with UCORT and OC. Conversely, the robust model revealed six significant associations that had been masked by outliers in the standard analysis (false negatives). Notably, robust positive correlations previously masked between CT and medial/overall muscle parameters (MHMCSA, MHCMI, and CCMI; P < 0.05), as well as between 25(OH)D and overall CCMI (P = 0.050), were successfully identified. Furthermore, hidden positive correlations between CTSMA-FF and UCORT and DR were also revealed (all P < 0.05), highlighting the fine sensitivity of intramuscular fat infiltration to systemic hormonal fluctuations following appropriate handling of extreme outliers.

Discussion

We did a study of acute changes in the components of the lower limbs in a microgravity environment. The relationship of these changes with biomarkers using a 7-day HDBR model found that during HDBR, muscle CSA of the lower limbs decreased, muscle contractile mass index of the lower limbs decreased, and subcutaneous fat decreased and then increased, and that there was a significant correlation between biomarkers such as UCORT, CORT, RBC, Ca, and P, and muscle/fat parameters, and that the lower limbs were bilaterally with asymmetry.

The rapid atrophy of thigh and gastrocnemius muscles during HDBR confirms the susceptibility of weight-bearing muscles to unloading^15,16. Our findings extend current knowledge by identifying integrated hormonal and metabolic pathways driving acute muscle loss.

25(OH)D levels are positively correlated with thigh muscle cross-sectional area (TMCSA), and vitamin D plays a role in regulating myogenic cell growth, size, and gene expression¹⁷. VDR knockout mice exhibit reduced grip strength and fibre atrophy due to myostatin upregulation¹⁸. These data support vitamin D supplementation as a potential countermeasure, as 1,25(OH)₂D directly inhibits MuRF1 transcription via VDR binding

In addition, changes in hormones such as growth hormone, cortisol, and growth factors, increased oxidative stress, and changes in neuromuscular function may also exacerbate the process of muscle atrophy¹⁹. DR correlated with reduced TMCSA, linking the Renin-angiotensin system (RAS) to local muscle pathology. Angiotensin II (Ang II) promotes oxidative stress via NAD(P)H oxidase²⁰, triggering mitochondrial dysfunction and protein breakdown²¹. Our 7-day HDBR protocol captures early atrophy linked to concurrent declines in 25(OH)D and rises in DR/UCORT. This highlights the acute sensitivity of muscle to integrated stress-metabolic perturbations, underscoring the need for multi-targeted interventions.

This study observed a significant reduction in CMI in the lower limbs of participants following 7 days of HDBR. This degenerative change reveals a dual tissue-physiological remodelling induced by early muscle unloading: on the one hand, the absence of mechanical stimulation leads to inhibited muscle protein synthesis and accelerated degradation, resulting in muscle fibre atrophy, manifested as a reduction in total muscle CSA²²; on the other hand, short-term disuse may result in ectopic fat deposition between muscle fibres or within cells⁷, manifested as an increase in PDFF²³. This phenomenon is consistent with the findings of Pagano et al., who demonstrated that after only 3 days in a dry immersion disuse model, subjects exhibited a significant decrease in muscle fiber cross-sectional area, and this short-term muscle deconditioning was accompanied by significant ectopic fat infiltration⁷. Compared with traditional assessment methods that rely solely on macroscopic morphology, CMI effectively eliminates the interference caused by hidden fat infiltration within the muscle. Consequently, even in an ultra-short-term disuse model lasting just 7 days, the decline in CMI can highly sensitively capture the loss of pure muscle tissue with genuine contractile function. This not only confirms that short-term weightlessness can induce substantial damage to skeletal muscle but also highlights the high sensitivity and superiority of the quantitative composite imaging marker CMI in the early prevention of space-induced weightlessness and the clinical monitoring of disuse-induced muscle atrophy.

The initial SAT reduction (BR3) followed by recovery contrasts with monotonic increases in long-term bed rest²⁴. Urinary cortisol (UCORT) inversely correlated with TSAT area, consistent with glucocorticoid-mediated activation of hormone-sensitive lipase²⁵. Altered cortisol levels in simulated microgravity may be psychologically stress-driven²⁶. DR negatively correlated with TSAT, as Ang II suppresses preadipocyte differentiation via AT1R²⁷. Our short-term data reveal a metabolic inflection point (BR3), where initial catabolism shifts to anabolism, likely driven by adaptive energy conservation. This challenges the assumption of unidirectional fat gain in the microgravity analogy.

Although this study provides valuable longitudinal data on lower extremity muscle and adipose changes during a 7-day-6° head-down bed rest (HDBR), three main limitations should be noted. Firstly, the sample size was relatively small and restricted to healthy adult males. Whilst this enhances internal homogeneity, it limits statistical power and the generalizability of the findings to broader populations. Secondly, the absence of a parallel ambulatory control group means that potential time-dependent or confined-environment effects cannot be entirely ruled out, although our longitudinal self-controlled design inherently mitigates many inter-individual confounders. Finally, from a methodological perspective, this study utilised the CSA of a single optimal slice for tissue quantification and employed CMI as a morphological proxy for muscle function. However, this approach has inherent limitations, including the lack of precise anatomical landmarks relative to the mid-femur and mid-tibia, the fact that both intra- and inter-rater reliability assessments were based on a relatively small sample size, and the potential for spatial displacement of the maximum CSA slice at follow-up, which may result in comparisons between different muscle regions. Future studies should incorporate three-dimensional whole-muscle volume analysis and direct biomechanical measurements to validate and extend these findings.

Although the findings of this study stem from controlled experiments involving healthy adult males, the core principle it reveals—that short-term strict immobilisation triggers significant muscular changes—holds cautionary and reference value for a broader population. It suggests that rapid disuse changes in the muscular system may be prevalent across diverse groups: astronauts undertaking short-term space missions; clinical patients requiring bed rest, particularly the elderly and those with chronic conditions; and individuals in occupations with restricted mobility. Consequently, early monitoring of muscle status and targeted interventions should become standard considerations for health maintenance across diverse real-world scenarios. Future research must validate and refine these findings in broader populations and more realistic models to advance cross-disciplinary strategies for muscle health preservation.

In conclusion, this study clarified the dynamic process of muscle-fat changes in short-term simulated microgravity, identified BR3 as a key intervention inflection point, and established the association between muscle atrophy and biomarkers such as vitamin D, RAS, and PTH, which will provide a new basis for astronauts’ personalized healthcare countermeasures.

Methods

Participants and Study Design

This study recruited healthy male volunteers through public advertising campaigns in the Beijing area. This prospective study was conducted from July 2024 to September 2024. This study was conducted in accordance with the principles of the Declaration of Helsinki and standard clinical practice guidelines. All protocols were approved by the Research Ethics Committee of Beijing Friendship Hospital, Capital Medical University (2024-P2-069-04), and were registered under ChiCTR2500096128. Prior to participation, all participants were provided with detailed explanations of the study procedures and potential risks, and written informed consent was obtained.

This study is an exploratory mechanism study. The sample size was determined primarily based on research feasibility and the common scale of similar head-down positioning bed studies. To ensure the detection of clinically meaningful differences and allow for potential dropouts, a final sample size of 40 healthy male participants was selected.

This study employed a non-randomized, self-controlled before-and-after experimental design. All volunteers meeting the predefined inclusion and exclusion criteria were assigned to receive identical intervention measures. The study utilized a non-randomized self-controlled before-and-after design at the individual participant level. Eligible volunteers were uniformly allocated to receive the same intervention protocol. To minimize potential biases, we implemented rigorous and standardized inclusion/exclusion criteria. A total of 44 healthy male participants were recruited for this study. Following comprehensive clinical interviews and physical assessments, participants were selected based on the following criteria. The inclusion criteria were defined as: (1) healthy male volunteers aged between 18 and 58 years; (2) height between 160 and 185 cm; (3) normal, regular baseline levels of physical activity; (4) normal results from comprehensive clinical and physical examinations. Exclusion criteria included: (1) a history of significant organic disease, particularly severe musculoskeletal, cardiovascular, hepatic, or renal disease; (2) hypertension, malnutrition, or metabolic/endocrine disorders; (3) a history of neurological, psychiatric, genetic, or infectious diseases ; (4) use of any medication known to affect bone or muscle metabolism within 6 months prior to enrolment; (5) presence of metallic implants, claustrophobia, or any other absolute contraindications to high-resolution MRI scanning.

The entire experimental procedure was conducted in a specialized, environment-controlled clinical research facility, including a 2-3 day pre-adaptation period, a rigorous 7-day-6° head-down positioning bed rest phase, followed by a subsequent 5-day recovery period. Prior to the commencement of the bed rest intervention, participants underwent an acclimatisation period to adapt to the standardised institutional diet and daily routine, ensuring physiological stability prior to baseline data collection. During the 7-day HDBR intervention, participants were strictly confined to bed, maintaining a continuous head-down position of -6° for 24 hours a day. To maintain the physiological validity of the disuse model, all daily activities (including eating, personal hygiene, and toileting) were strictly performed in the -6° position, with no weight-bearing activities or head elevation permitted during this period; compliance was strictly monitored by specially trained medical staff via 24-hour video surveillance. Following the bedrest phase, participants resumed an upright posture under supervision and gradually returned to normal walking activity during a 5-day recovery period. To dynamically track the longitudinal trajectories of lower limb morphological changes and systemic biochemical fluctuations, comprehensive data collection was conducted at five predetermined time points: baseline (PRE, collected immediately at the end of the pre-adaptation period and prior to the start of HDBR bed rest), day 3 of bed rest (BR3), day 7 of bed rest (BR7), day 3 of recovery (R3), and day 5 of recovery (R5). At each of these time points, participants underwent MRI scans of both thighs and calves, and fasting venous blood samples were collected in the early morning to assess systemic biomarker profiles(Fig. 6).

Fig. 6. Experimental flow chart.

MRI were taken at baseline, BR3, BR7, R3, and R5 (Day 3 and 7 head-down tilt bed rest, Day 3 and 5 recovery, respectively).

Dietary Control

In order to strictly control for nutritional confounding factors, a standardised dietary regimen was strictly adhered to during all three experimental phases (the pre-adaptation phase, the HDBR phase, and the recovery phase). The diets of the participants in our trial were based on the Dietary Guidelines for Chinese Residents (2022) and the Chinese Dietary Reference Intakes (DRIs) (2023 Revision). The dietary plan also incorporated the nutritional support program of the previous HDBR study²⁸. The average daily calorie intake was approximately 2800 kcal and varied by ≤10% over a 3-day menu cycle.

MRI Measurements

MRI examinations were performed on a 3.0 T scanner (Siemens Vida, Erlangen, Germany). Participants were placed in the supine position, and a combination of anterior and posterior coil arrays was used. Participants were scanned with MRI at five time points (PRE, BR3, BR7, R3, and R5). For each participant at each time point, thigh and calf images were acquired separately, (T1-weighted 2-point Dixon sequence, slice thickness = 2.0 mm; distance factor = 20%, TR = 4.34 ms, TE1 = 1.34 ms, TE2 = 2.76 ms, flip angle = 9 deg; field of view = 400 mm × 324 mm at 1.3 mm × 1.3 mm pixel size). Total scan time was reduced to a minimum (total scan time of 5 to 7 min) and embedded in the patient’s regular clinical follow-up schedule.

Biochemical biomarkers Collection

All blood samples were collected from the antecubital vein in the early morning, after the subjects had fasted for at least 8 hours, at the following five study time points: baseline, day 3 of bed rest, day 7 of bed rest, day 3 of recovery, and day 5 of recovery. Participants were instructed to avoid caffeine, alcohol, and strenuous exercise for at least 24 hours prior to each sampling. Urine samples were collected simultaneously in the early morning on the same day as each blood sample collection, using a clean midstream urine sample. The serum biochemical parameters measured in this study included: calcium (Ca), phosphorus (P), direct renin (DR), adrenocorticotropic hormone (ACTH), intact parathyroid hormone (IPTH), calcitonin (CT), osteocalcin (OC), 25-hydroxyvitamin D [25(OH)D], red blood cell count (RBC), white blood cell count (WBC), haemoglobin (Hb), and cortisol (CORT); the urinary parameter was urinary cortisol (UCORT). All tests were performed in the hospital’s accredited central laboratory. Hormonal parameters were measured using validated chemiluminescent immunoassays; routine blood parameters were analysed by a fully automated haematology analyser; and electrolyte parameters were measured by a clinical biochemistry analysis system. All tests included internal quality control, and laboratory personnel were blinded to the radiological findings. Where feasible, measurements were repeated, and the mean value was taken. All results were reported in the International System of Units (SI) as provided by the laboratory.

Skeletal Muscle Segmentation

In the MRI PDFF axial images, we examined the entire length of the thigh and calf muscle groups layer by layer, and independently selected the axial slice with the largest cross-sectional area within the mid-diaphysis of the femur and tibia as the measurement slice. Once the slice had been selected, the contours of the target muscle groups, bone marrow and subcutaneous fat are manually traced using ITK-SNAP²⁹ (www.itksnap.org, accessed on 3 February 2020). The software automatically calculates the cross-sectional area and fat fraction of the outlined regions. To ensure the reliability of the measurements, all segmentation results were reviewed by a certified radiologist with 19 years’ experience in whole-body imaging. Representative results are shown in Fig. 7. To evaluate the reliability of manual segmentation, two independent analyses were conducted. For intra-rater reliability, the same analyst reanalyzed a random selection of 10 thigh and 10 calf MRI images after an interval of several months, blinded to the initial results. For inter-rater reliability, a second independent analyst performed blinded segmentation and measurements on a separate, randomly selected set of 10 thigh and 10 calf MRI images. Measurement consistency was assessed by calculating the intraclass correlation coefficient (ICC). Segmentation was performed by radiologists unaware of scan timestamps to minimize measurement bias.

Fig. 7. Segmentation of thighs and calves.

A Segmentation of the right and left thigh muscle cross-sectional area (red and green), thigh subcutaneous adipose tissue area (blue and yellow), and thigh bone marrow adipose tissue fat fraction (light blue and purple). B Segmentation of the right and left calf medial head of the gastrocnemius muscle cross-sectional area (red and yellow), lateral head of the gastrocnemius muscle cross-sectional area (green and light blue), calf subcutaneous adipose tissue area (blue and purple), calf bone marrow adipose tissue fat fraction (apricot and dark blue).

CMI Measurements

Based on the extracted PDFF and cross-sectional area (CSA) values, the contractile mass index (CMI) was calculated as follows: $\mathit{CMI}=\mathit{CSA}\times (1-\mathit{PDFF})$³⁰.

Statistical Analysis

The analytical unit of this study was individual participants. All statistical analyses were performed based on data at the individual level. The normality of continuous variables was assessed using the Shapiro-Wilk test. Data are presented as mean ± standard deviation. To minimise measurement bias and reflect the overall anatomical condition, the MRI measurements from both sides (left and right) of each subject were averaged prior to formal statistical modelling for subsequent analysis.

To compare the differences in rates of change between the thigh and calf, the percentage change in each MRI parameter from baseline (PRE) to day 7 of bed rest (BR7) was calculated using the following formula: Rate of change Δ = [(BR7 measurement − PRE measurement)/PRE measurement] × 100%. Differences in the rates of change between the two sites were analysed using a paired t-test (for normally distributed data) or the Wilcoxon signed-rank test (for non-normally distributed data).

To facilitate direct comparison of effect sizes between variables with different units of measurement, and to ensure the numerical stability and optimal convergence of the linear mixed model, all continuous longitudinal data were standardised using Z-scores prior to formal modelling. A linear mixed model (LMM) was employed to analyse the longitudinal changes in MRI measures across five time points (PRE, BR3, BR7, R3, and R5). The Z-score-standardised MRI measures were set as the dependent variable. Time was treated as a fixed effect, and participant ID as a random intercept. Post-hoc pairwise comparisons between time points were corrected using the Bonferroni method. A separate LMM was constructed to assess the longitudinal association between blood biochemical markers and MRI measures. Z-score-standardised MRI measures were set as the dependent variable, individual Z-score-standardised biochemical marker measures were set as fixed effects, and participant ID was set as a random intercept. Effect sizes were reported as standardised regression coefficients β. To control for the influence of potential outliers, a robust linear mixed model (Robust LMM) was employed for sensitivity analysis. Associations that were statistically significant in both the standard LMM and the robust LMM were considered final significant results.

Intra-rater reliability and inter-rater reliability for the MRI measurements were assessed using the Intraclass Correlation Coefficient (ICC) with a two-way mixed-effects model. An ICC > 0.75 was considered to indicate excellent agreement. A two-sided P < 0.05 was taken as indicating a statistically significant difference. Data analysis was performed using R version 4.4.2.

Author contributions

Y.S.: Original draft preparation, MRI measurements, and multi-tissue segmentation (including skeletal muscle, subcutaneous adipose tissue, and bone marrow), W.Y.: Data analysis, multi-tissue segmentation (including skeletal muscle, subcutaneous adipose tissue, and bone marrow), and original draft preparation, Y.L.: Data analysis, Z.W.: Multi-tissue segmentation (including skeletal muscle, subcutaneous adipose tissue, and bone marrow), Y.L., R.W., H.H., L.C., Y.H., and P.Q.: MRI measurements, P.R.: Manuscript reviewing and modifications, Z.W.: Conceived and designed this study.

Data availability

The datasets generated and/or analyzed during the current study are not publicly available due to privacy and ethical restrictions, but are available from the corresponding author on reasonable request.

Code availability

The code used for analysis in this manuscript will be made available. The code for this study is available in Figshare and can be accessed via this link https://doi.org/10.6084/m9.figshare.31938558.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41526-026-00640-x.