Reference values and cut-points for trunk myopenia and myosteatosis in Chinese adults: a secondary analysis of the China Action on Spine and Hip study Authors: Ling Wang, Fangfang Duan, Yandong Liu, Dong Yan, Kai Li, Wenkai Wu, Yi Yuan, Quanzhong Ren, Kangkang Ma, Fengyun Zhou, Zitong Cheng, Jian Geng, Renxian Wang, Xiaoguang Cheng, Wenshuang Zhang Published: 2026 ABSTRACT BackgroundSarcopenia is a prevalent condition associated with adverse health outcomes, necessitating accurate assessment of skeletal muscle quantity and quality. Computed tomography (CT) is the gold-standard imaging modality for evaluating trunk skeletal muscle area (TrSMA) and density (TrSMD), with opportunistic CT offering a means to assess these metrics without additional radiation exposure. However, the clinical application of CT-based sarcopenia assessment in Chinese populations is hindered by the absence of large-scale, population-specific reference values and cut-off points, particularly for both myopenia (low muscle mass) and myosteatosis (fatty infiltration). This study aims to establish sex- and age-specific percentile reference values and diagnostic cut-off points for CT-derived TrSMA and TrSMD at the L1 and L3 vertebral levels with a large cohort of community-dwelling Chinese adults.MethodsWe retrospectively analyzed abdominal CT scans from 4,016 community‑dwelling Chinese adults (21–80 years) across 12 regions. TrSMA and TrSMD at L1 and L3 were segmented using OsiriX. Sex- and age-specific percentile curves were modeled using the lambda-mu-sigma (LMS) method within the generalized additive models for location, scale, and shape (GAMLSS) framework. Estimated age-related differences in TrSMA and TrSMD from ages 25 to 75 years were calculated using LMS-derived medians. Sex‑specific cut‑off points for myopenia and myosteatosis were defined in the 21- to 40-year subgroup using the mean minus two standard deviations (M‑2SD) approach.ResultsFrom ages 25 to 75 years, LMS-derived L1/L3-TrSMA values were lower by 0.45%/0.55% per year in men and 0.33%/0.41% per year in women; corresponding differences in L1/L3-TrSMD were 0.38%/0.35% and 0.55%/0.52% per year. More pronounced age-related decreases were observed in peri- and postmenopausal women and older men. M‑2SD cut‑offs at L1 were 96.0 cm2 and 29.8 Hounsfield Unit (HU) (men), 63.1 cm2 and 26.4 HU (women); at L3 were 122.3 cm2 and 31.7 HU (men), 75.4 cm2 and 27.5 HU (women).ConclusionsWe provide the first CT‑based, sex‑ and age‑specific reference values and cut‑points for trunk myopenia and myosteatosis in Chinese adults, facilitating standardized sarcopenia diagnosis and research. Introduction Sarcopenia, a condition marked by diminished skeletal muscle function, strength, and mass, presents a substantial health challenge due to its multifaceted origins (1). Sarcopenia has severe consequences that include physical impairment, a decline in life quality, and an increased risk of death (1,2). Accurate diagnosis of sarcopenia depends on measuring muscle quantity as well as quality. To tackle this, a range of medical imaging modalities, such as computed tomography (CT), magnetic resonance imaging (MRI), and dual-energy X-ray absorptiometry (DXA), is essential in defining the muscle characteristics linked to this illness. CT and MRI are the gold standards for the non-invasive assessment of body composition, playing a crucial role in evaluating trunk muscle mass (3,4). However, all these imaging modalities have limited availability for screening for sarcopenia (5). One potential solution is to use opportunistic abdominal CT images to assess the trunk skeletal muscle area (TrSMA; in cm2) and density (TrSMD; in Hounsfield Units, HU) without exposing patients to additional radiation (6,7). Sites at the levels of the first (L1) or third lumbar (L3) vertebrae are frequently chosen because they provide important information on the properties of the trunk musculature, and the L3 level is recommended by the European Working Group on Sarcopenia in Older People 2 (EWGSOP2) (5,8,9). CT image-based surrogate markers of sarcopenia are currently recognized as belonging to two distinct types, namely, “myopenia” and “myosteatosis” (10,11). While sarcopenia is defined as the loss of muscle strength, mass, and function, myopenia refers to low skeletal muscle mass and is measured by using CT at the level of L3 to measure surrogates of TrSMA or by using DXA to measure appendicular lean mass (ALM). The trunk skeletal muscle index (TrSMI), calculated by adjusting TrSMA by the patient’s height [TrSMI = TrSMA/height2 (cm2/m2)], is a frequently used and more effective diagnostic tool than TrSMA alone for identifying myopenia by providing an accurate assessment of muscle mass in relation to body size. Myosteatosis is defined as the infiltration of fat into skeletal muscle, identifiable through reduced muscle attenuation on CT images (12). It can be quantified by using measures such as TrSMD, which can change independently of muscle mass and may synergistically affect muscle function (13). While CT scans have been used extensively to measure trunk muscle area, density, and index in Asian cohorts, especially in Korea and China, most of these studies have been conducted on subsets such as elderly persons or individuals with particular medical disorders, which has limited their wider application to the general population. Significant differences in the cutoff values for muscle measurements have been found in investigations conducted on healthy populations both within and between Asian nations. For example, cutoff values are usually computed as the mean minus two standard deviations (M-2SD) using data for young adults without significant illnesses. In Korea, L3-TrSMI cutoff points range from 39.8 to 42.9 cm2/m2 for males and 28.5 to 31.0 cm2/m2 for females (14,15), while in northern China they are set at 37.9 cm2/m2 for males and 28.6 cm2/m2 for females (16). More importantly, there is a lack of multicenter studies focused on trunk muscle myopenia and myosteatosis in the adult population in China. This research gap limits our ability to precisely define sarcopenia in Asian or Chinese populations using CT imaging. The primary objective of this study is to utilize opportunistic abdominal CT data from a large sample of community-dwelling adult population across China to measure the trunk muscle area and density at the L1 and L3 levels, and the secondary objective is to establish corresponding reference values. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2026-0751/rc). Methods Study population In this retrospective analysis, we used the baseline CT images from the prospective China Action on Spine and Hip (CASH) cohort study (ClinicalTrials.gov Identifier: NCT01758770) (6). This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The CASH study used multi-stage cluster sampling and collected baseline data and abdominal CT scans between 2013 and 2017. Sampling was performed at the region, community or village, and household levels. Urban areas were defined by neighborhoods or postal codes, rural by villages. Households qualified if residents lived continuously ≥4 years. All participants provided written informed consent. Pregnant individuals and those with metallic implants or limited mobility were excluded. The study initially recruited 4,305 adults aged 21 to 80 years, providing free abdominal CT scans and baseline data collection. Participants were excluded due to poor-quality or incomplete CT images (n=127), insufficient or inconsistent baseline data (n=76), spinal abnormalities (n=32), ankylosing spondylitis (n=1), or prior spine or hip fixation surgery (n=53). Ultimately, 4,016 participants (1,524 men, 2,492 women) were included. Age, sex, weight, height, waist, and hip circumferences were recorded. Body mass index (BMI) and waist-hip ratio (WHR) were calculated as BMI = weight/height2 (kg/m2) and WHR = waist circumference/hip circumference. CT examination An abdominal CT scan was conducted with a lateral scout targeting the T4 to L4 vertebrae for localization. The scanning protocols were predetermined and tailored to each facility. The scanning process consistently employed a 120 kVp setting, while other scanning parameters varied between centers based on the specific CT imaging equipment used (6). To ensure cross-calibration consistency, a European Spine Phantom (ESP-145, QRM GmbH, Möhrendorf, Germany) was scanned at all participating centers prior to any subjects (17). Quality assurance analyses showed minimal cross-center variation in the water-equivalent CT value of the ESP, with an average difference of less than 2.6 HU (18). Muscle area and density measurement Measurements of paraspinal, posterior abdominal, and anterolateral abdominal muscles were performed at L1 and L3 levels (Figure 1) using OsiriX software (Lite 10.0.2, Pixmeo, Geneva, Switzerland). Muscle segmentation was conducted semiautomatically with the pencil tool and standardized thresholds (−29 to 150 HU) to calculate TrSMA and TrSMD. A lower threshold of −29 HU separated muscle from adipose tissue. Measurements were obtained by five radiologists with ≥5 years of experience. Interobserver reliability was assessed with 50 images analyzed by two radiologists, and intraobserver reliability was evaluated by repeat measurement of 30 images one month later. Data underwent cleaning to exclude outliers beyond ± 3SD, reviewed and corrected by a senior radiologist before final analysis. Schematic diagram of trunk muscle measurement at the L1 and L3 levels based on abdominal computed tomography. (A) Reconstructed sagittal CT images of the lumbar spine region for positioning at the L1 and L3 levels, with two red lines indicating the planes corresponding to L1 and L3 levels, respectively; (B) and (C) are diagrams illustrating the delineation of the ROIs for trunk muscles at the L1 and L3 levels, where the purple dashed lines represent the delineation path, and the green filled areas represent the TrSMA. CT, computed tomography; ROIs, regions of interest; TrSMA, trunk skeletal muscle area. Statistical analysis The normality of continuous variables was assessed using Q-Q plots. Variables with no marked curvature or tail deviation were considered approximately normally distributed and reported as mean ± standard deviation (SD). Other variables were reported as median (interquartile range, IQR). The consistency of CT scan interpretations between and within observers was quantified using the intraclass correlation coefficient (ICC). Statistical analysis was performed with SPSS 22.0 (IBM Corp., Armonk, NY, USA). Gender differences in continuous data were compared using t-tests or Mann-Whitney rank tests, considering a P value less than 0.05 as statistically significant. Age-specific percentile reference values were estimated using the lambda-mu-sigma (LMS) method, incorporating the median (M), the coefficient of variation (S), and the Box-Cox transformation power (L) (19). Age was modeled as a continuous variable in the LMS analysis, without prior grouping into 5- or 10-year age bands. The L, M, and S parameters were allowed to vary smoothly with age within the generalized additive models for location, scale, and shape (GAMLSS) framework. Using R version 4.2.3 and the GAMLSS package (version 5.4-12), we estimated sex-specific percentiles (P3, P10, P25, P50, P75, P90, P97) for TrSMA and TrSMD at the L1 and L3 levels and generated the percentile curves. Model fitting involved the Normal distribution (NO), Box-Cox-t (BCT), Box-Cox Cole and Green (BCCG), and Box-Cox Power Exponential (BCPE) distributions. The optimal model was chosen based on the Generalized Akaike Information Criterion (GAIC) (20). The 5-year values presented in the tables were extracted from the fitted curves for reporting purposes. Standardized age-related differences were calculated from LMS-derived median values and interpreted as cross-sectional patterns, not longitudinal within-person change. The M-2SD method, based on the 21- to 40-year age group, was used to define the reference-based cut-off points for trunk muscle metrics (16). As a sensitivity check, non-parametric thresholds were also calculated in the young reference group using the fifth lowest value in men and the sixth lowest value in women. The standard error and 95% CI of each M-2SD cut-off were estimated by combining the standard errors of the mean and standard deviation. Results The study included 4,016 abdominal CT scans of adults aged 21 to 80 years (2,492 women, 1,524 men). The mean ages of the females and males were 57.3±12.3 and 57.4±13.4 years, and their mean BMIs were 24.6±3.6 and 25.0±3.4 kg/m2, respectively. Table 1 presents the general characteristics of the women and men in 20-year age groups. Their general characteristics in 5-year age groups are presented in Table S1. The reproducibility analysis showed good agreement between the image segmentation measurements, with high intra-observer (ICC =0.981–0.989, all P<0.001) and inter-observer (ICC =0.984–0.991, all P<0.001) agreement (Table S2). Table 1Age- and sex-specific distribution of general characteristics parameters, grouped by 20-year age bandsParametersYoung (21–40 years) (n=436)Middle-aged (41–60 years) (n=1,674)Elderly (61–80 years) (n=1,906)Male (n=199)Female (n=237)PMale (n=567)Female (n=1,107)PMale (n=758)Female (n=1,148)PAge (years)33.1±4.632.2±4.80.05751.3±5.751.8±5.50.08768.4±5.267.8±5.00.016Height (cm)172.5±6.2161.2±5.6<0.001168.0±7.0157.5±6.0<0.001166.1±6.7155.3±5.8<0.001Weight (kg)77.5±12.860.6±9.8<0.00171.2±11.561.0±9.8<0.00168.1±11.460.1±9.7<0.001BMI (kg/m2)26.0±3.723.4±3.7<0.00125.2±3.324.6±3.60.00124.6±3.324.9±3.60.096Waistline (cm)89.4±9.477.7±8.8<0.00186.8±10.080.9±10.1<0.00186.1±10.183.2±9.7<0.001Hipline (cm)100.7±6.695.5±6.5<0.00196.7±7.595.2±7.9<0.00196.1±8.996.1±9.10.885WHR0.89±0.060.81±0.06<0.0010.90±0.060.85±0.07<0.0010.90±0.070.87±0.07<0.001Data are shown as mean ± standard deviation. P values indicate differences between women and men within each age band. All continuous variables were approximately normally distributed based on Q-Q plots and were compared using independent-samples t-tests. P<0.05 was considered statistically significant. BMI, body mass index; WHR, waist-hip ratio. Age- and sex-specific distribution of general characteristics parameters, grouped by 20-year age bands Data are shown as mean ± standard deviation. P values indicate differences between women and men within each age band. All continuous variables were approximately normally distributed based on Q-Q plots and were compared using independent-samples t-tests. P<0.05 was considered statistically significant. BMI, body mass index; WHR, waist-hip ratio. LMS-derived fitted medians (P50) and quartiles (P25 and P75) of trunk muscle parameters between 25 and 75 years of age in 5-year intervals are shown in Tables S3,S4. In general, men exhibited greater muscle area (L1-TrSMA and L3-TrSMA) and density (L1-TrSMD and L3-TrSMD) than women (Table S5, all P<0.001). Differences in trunk muscle parameters at the L1 and L3 levels across 5-year age intervals were expressed as standardized age-related differences per year (%/year), relative to the LMS-derived median value at age 25 years (Table 2). The standardized age-related difference was calculated using Eq. [1], as shown below: Table 2Standardized age-related differences in LMS-derived trunk muscle metrics across 5- and 50-year age intervals (% per year)Age interval (years)MaleFemaleL1-TrSMA (cm2)L1-TrSMD (HU)L3-TrSMA (cm2)L3-TrSMD (HU)L1-TrSMA (cm2)L1-TrSMD (HU)L3-TrSMA (cm2)L3-TrSMD (HU)25 to <30−0.26% −0.34%−0.31%−0.21%−0.19%−0.54%−0.34%−0.50%30 to <35−0.31%−0.32%−0.41%−0.26%−0.23%−0.54%−0.35%−0.45%35 to <40−0.42%−0.28%−0.56%†−0.23%−0.23%−0.54%−0.36%−0.30%40 to <45−0.41%−0.20%−0.57%†−0.13%−0.24%−0.54%−0.38%−0.27%45 to <50−0.34%−0.20%−0.51%−0.19%−0.34%†−0.55%†−0.41%−0.47%50 to <55−0.30%−0.24%−0.48%−0.26%−0.42%†−0.56%†−0.43%†−0.63%†55 to <60−0.35%−0.31%−0.46%−0.32%−0.43%†−0.55%†−0.45%†−0.58%†60 to <65−0.45%−0.44%†−0.50%−0.47%†−0.40%†−0.55%†−0.46%†−0.60%†65 to <70−0.78%†−0.63%†−0.75%†−0.67%†−0.39%†−0.55%†−0.46%†−0.72%†70 to 75−0.93%†−0.80%†−0.91%†−0.79%†−0.40%†−0.54%−0.47%†−0.74%†25 to 75−0.45%−0.38%−0.55%−0.35%−0.33%−0.55%−0.41%−0.52%Values represent standardized age-related differences between LMS-derived medians across each 5-year age interval. Differences were expressed as percent difference per year relative to the LMS-derived medians at age 25 years. The final row shows the overall average standardized age-related difference from ages 25 to 75 years. These values were derived from the LMS-derived medians presented in Tables S3,S4. Values marked with † indicate age intervals with more pronounced age-related decreases than the overall 25- to 75-year average for the corresponding metric. HU, Hounsfield Unit; LMS, lambda-mu-sigma; TrSMA, trunk skeletal muscle area; TrSMD, trunk skeletal muscle density. Standardized age-related differences in LMS-derived trunk muscle metrics across 5- and 50-year age intervals (% per year) Values represent standardized age-related differences between LMS-derived medians across each 5-year age interval. Differences were expressed as percent difference per year relative to the LMS-derived medians at age 25 years. The final row shows the overall average standardized age-related difference from ages 25 to 75 years. These values were derived from the LMS-derived medians presented in Tables S3,S4. Values marked with † indicate age intervals with more pronounced age-related decreases than the overall 25- to 75-year average for the corresponding metric. HU, Hounsfield Unit; LMS, lambda-mu-sigma; TrSMA, trunk skeletal muscle area; TrSMD, trunk skeletal muscle density. where Rate is the standardized age-related difference per year, expressed as a percentage and standardized relative to the LMS-derived median value at age 25; n represents age; n1–n2 represents the age range from age n1 to n2; f(n) refers to the LMS-derived medians of trunk skeletal muscle metrics, such as muscle area or muscle density, at the L1 or L3 levels for individuals aged n; for example, f(25) specifically denotes the LMS-derived medians of trunk skeletal muscle metrics at age 25 years. Age intervals with more pronounced age-related decreases than the overall 25- to 75-year average are shown in Table 2. Based on the sex-specific percentile values (P3, P10, P25, P50, P75, P90, P97) of TrSMA and TrSMD across different ages at the L1 and L3 levels, the fitted percentile curves are illustrated in Figures 2,3, respectively. The LMS-derived P3 and P97 of trunk skeletal muscle metrics for individuals aged 25 to 75 years, at 5-year intervals, are also detailed in Table 3. Percentile charts of L1-TrSMA (A,B) and L3-TrSMA (C,D) in males and females aged 21 to 80 years. TrSMA, trunk skeletal muscle area. Percentile charts of L1-TrSMD (A,B) and L3-TrSMD (C,D) in males and females aged 21 to 80 years. HU, Hounsfield Unit; TrSMD, trunk skeletal muscle density. Table 3The LMS-derived P3 and P97 of trunk skeletal muscle metrics for individuals aged 25–75 years (with 5-year interval)Age (years)MaleFemaleL1-TrSMA (cm2)L1-TrSMD (HU)L3-TrSMA (cm2)L3-TrSMD (HU)L1-TrSMA (cm2)L1-TrSMD (HU)L3-TrSMA (cm2)L3-TrSMD (HU)P3 2597.3731.62122.6734.1669.0429.6784.3830.56 3096.0931.93120.7533.7568.6928.9782.9529.52 3594.5831.76118.2233.0168.2328.2681.4828.84 4092.5431.09114.8132.1367.5127.5179.9628.72 4590.5430.25111.2931.3566.3226.6878.3628.65 5088.9029.67108.1830.8864.3425.7276.6427.91 5587.4329.54105.2331.0362.0624.6574.8026.67 6085.7429.25102.4330.9860.4323.4972.9025.32 6583.5528.4499.3630.1559.4122.2570.9823.85 7079.7726.9394.7628.2758.2020.9369.0322.01 7575.2324.9689.1925.7356.0119.5667.0619.72P97 25177.7054.43216.5954.39121.9852.54147.5053.82 30175.3653.25213.2053.81120.3051.03144.9852.74 35172.6052.18208.7353.01118.4149.53142.4251.54 40168.8751.41202.7152.32116.9348.07139.7650.40 45165.2250.96196.4952.01116.1146.67136.9749.34 50162.2350.22191.0151.58115.3545.35133.9648.11 55159.5449.20185.8050.86114.1944.11130.7446.71 60156.4748.25180.8550.14111.7643.00127.4345.62 65152.4847.32175.4349.27108.6542.01124.0644.57 70145.5746.20167.3048.13105.9841.09120.6643.40 75137.2944.84157.4746.68104.9840.25117.2142.56HU, Hounsfield Unit; LMS, lambda-mu-sigma; P3, 3rd percentile; P97, 97th percentile; TrSMA, trunk skeletal muscle area; TrSMD, trunk skeletal muscle density. The LMS-derived P3 and P97 of trunk skeletal muscle metrics for individuals aged 25–75 years (with 5-year interval) HU, Hounsfield Unit; LMS, lambda-mu-sigma; P3, 3rd percentile; P97, 97th percentile; TrSMA, trunk skeletal muscle area; TrSMD, trunk skeletal muscle density. The percentile curves in Figures 2,3 illustrate age-related patterns of trunk muscle area and density, with lower LMS-derived median values observed at older ages. As shown in Table 2, between the ages of 25 and 75 years, men displayed more pronounced age-related decreases in trunk muscle area (L1-TrSMA, −0.45%/year; L3-TrSMA, −0.55%/year) than women (L1-TrSMA, −0.33%/year; L3-TrSMA, −0.41%/year), while women exhibited greater age-related declines in trunk muscle density (L1-TrSMD, −0.55%/year and L3-TrSMD, −0.52%/year) than men (L1-TrSMD, −0.38%/year and L3-TrSMD, −0.35%/year). Furthermore, age-related decreases in TrSMA were more pronounced at the L3 level than the L1 level, whereas the decreases in TrSMD at the L1 and L3 levels were comparable. A more pronounced age-related decrease is defined as a standardized age-related decrease within a 5-year interval that exceeded the overall average decrease across the 25–75-year age range. For men in midlife (35–45 years) and old age (65–75 years), more pronounced age-related decreases were observed in L3-TrSMA and in L1-TrSMA for the old age group (65–75 years). Furthermore, more pronounced age-related decreases in L1-TrSMD and L3-TrSMD were also observed in older men aged 60–75 years. Among peri- and postmenopausal women aged 45–75 years, more pronounced age-related decreases were observed in both trunk muscle area and density. Notably, TrSMA and TrSMD at the L1 and L3 levels both showed the largest age-related decreases at age 70–75 years in men and L3-TrSMD at age 70–75 years in women. Specifically, the standardized age-related decline in L1-TrSMD in women was relatively stable from ages 25 to 75 years, with an average standardized age-related decrease of −0.55%/year (range, −0.56%/year to −0.54%/year). The cut-off points for TrSMA, TrSMD, and TrSMI determined using the M-2SD threshold in younger adults (ages 21–40 years) were as follows: at the L1 level, 96.0 cm2, 29.8 HU, and 32.5 cm2/m2 for men, and 63.1 cm2, 26.4 HU, and 24.5 cm2/m2 for women; at the L3 level, 122.3 cm2, 31.7 HU, and 41.3 cm2/m2 for men, and 75.4 cm2, 27.5 HU, and 29.4 cm2/m2 for women. For TrSMA and TrSMD, non-parametric thresholds based on the fifth lowest value in men and the sixth lowest value in women were generally comparable with the M-2SD thresholds (Table S6). The standard errors and 95% CIs of the M−2SD cut-offs for TrSMA and TrSMD are shown in Table S6. Discussion To our knowledge, this is the first study using the LMS method to generate sex- and age-specific percentile reference values for trunk muscle area and density among a large cohort of Chinese community-dwelling adults. These reference values facilitate the evaluation of trunk muscle myosteatosis and myopenia in adult Chinese populations, supporting clinical diagnosis and epidemiological research. Skeletal muscle is vital for functional health, particularly among older adults; thus, establishing age-specific reference values is essential. Previously, our group created percentile charts for sex-specific CT-derived paravertebral muscle metrics in Chinese populations (21), highlighting distinct degeneration patterns between sexes. Expanding upon this, our study analyzed CT-derived metrics at L1 and L3 levels, generating tailored percentile charts for individuals aged 25 to 75 years. We also established sex-specific CT-derived cut-off points using mean minus two standard deviations for younger adults (21 to 40 years). Our findings demonstrate age-related declines in trunk muscle area, density, and index, confirming both muscle volume reduction (myopenia) and increased fat infiltration (myosteatosis). These observations underscore the need for evaluating muscle quantity and quality concurrently, aligning with previous research (16,22). We observed significant sex differences in the age-related decreases in trunk muscle metrics. Men exhibited more pronounced age-related decreases in trunk muscle area at L1 and L3 compared to muscle density, whereas women showed more pronounced age-related decreases in muscle density. Within the same age group, female trunk muscle density at L1 and L3 was consistently lower than that of males, in agreement with earlier findings (23,24). These gender differences likely reflect physiological factors, hormonal influences, and lifestyle variations (22,25). Reduced estrogen during menopause may notably influence muscle density and area decline in women, promoting lipid accumulation and adipogenic gene activation within skeletal muscle (26). Consequently, the pronounced decrease in muscle density among females indicates more severe muscle quality deterioration, as lower density suggests greater fat infiltration (27), linked to metabolic dysfunctions, inflammation, and reduced muscle function (28). Nevertheless, a common finding in both sexes was that trunk muscle area declined more rapidly at the L3 level than at L1, while trunk muscle density reductions remained consistent between these levels. This indicates more pronounced volume reduction at L3 but uniform fat infiltration across spinal levels. This variability suggests complex physiological mechanisms behind age-related muscle degeneration. Greater myopenia at L3 might result from differential muscle functions, stress conditions, and muscle atrophy rates between spinal levels (22-24). Conversely, uniform myosteatosis across spinal levels could reflect systemic metabolic alterations common in aging (28,29). Future studies should explore the physiological basis for these variations and potential targeted interventions to mitigate muscle degeneration. Analysis of standardized age-related decreases across 5-year intervals showed more pronounced declines in trunk muscle area at L3 among men aged 35 to 45 years, potentially related to midlife weight gain (30), hormonal fluctuations (31), dietary changes (25,32), or reduced physical activity (32). However, no corresponding increase in muscle density decline was observed in this group. Among women, L1 trunk muscle density consistently declined at rates between −0.54% and −0.56% per year across age groups, indicating a steady progression of structural and functional muscle deterioration with aging. Few studies have established sex-specific CT-derived cut-off points for trunk muscle metrics among Chinese populations. Ming Kong et al. (16) previously proposed trunk muscle area and index cut-off points at L3 using the 5th percentile and mean minus two standard deviations methods, but did not include muscle density, a key quality metric. Compared to their study, our TrSMI cut-off points at L3 for adults aged 20 to 39 years were slightly higher, likely due to higher BMI in our cohort. Our methodological approach enhanced demographic representativeness through multi-stage cluster sampling across 12 regions, encompassing both urban and rural populations and minimizing selection bias. Conversely, Kong et al.’s study relied on retrospective outpatient data without significant catabolic diseases, potentially limiting applicability. Table 4 summarizes global studies reporting CT-derived trunk muscle cut-off points at L3 or L1 for healthy populations, facilitating comparisons. Table 4Studies establishing CT-derived cut-off points at L3 or L1 level to evaluate trunk muscle metrics among healthy individualsNo.AuthorYearCountrySoftware (threshold)Study populationIndividual numbersAge (years)SiteParametersMethodsCut-off pointsMaleFemale1Kong et al. (16)2022ChinaSliceOmatic (−29 to 150 HU)Outpatient patients without serious illnessn=700 (M/F =363/337)20–40L3TrSMIP540.231.6M-2SD37.928.6n=1,426 (M/F =723/703)20–60L3TrSMIP540.031.3M-2SD3728.82Kim et al. (14)†2021KoreaAn automated software (−29 to 150 HU)Routine health check-up participantsN=1,917 (M/F =1,222/695)20–44L3SMAM-2SD119.374.2SMIM-2SD39.828.53Kim et al. (15)†2021KoreaAn automated software (−29 to 150 HU)Routine health check-up participantsN=3,260 (M/F =2,100/1,160)20–44L3SMAM-2SD128.481.2SMDM-2SD40.239.9SMIM-2SD42.931.04Bahat et al. (33)2021TurkeyExtreme PACS (−30 to 150 HU)Healthy liver donorsn=482 (M/F =268/214)18–40L3TrSMAM-2SD121.269.5P5132.083.0TrSMIM-2SD38.727.8P541.430.7n=601 (M/F =326/275)18–60L3TrSMAM-2SD118.970.2P5131.384.0TrSMIM-2SD37.827.8P541.331.4n=119 (M/F =58/61)41–60L3TrSMIM-2SD34.328.1P540.932.15van der Werf et al. (34)2018The NetherlandsSliceOmatic (−29 to 150 HU)Healthy kidney donorsn=300 (M/F =126/174)20–60L3TrSMAP5138.296.2TrSMDP530.924.8TrSMIP543.132.7n=420 (M/F =174/246)20–82L3TrSMAP5134.089.2TrSMDP529.322.0TrSMIP541.632.06Derstine et al. (35)2018USAMatlab (−29–150 HU)Healthy kidney donorsn=727 (M/F =317/410)18–40L3TrSMAM-2SD144.392.2TrSMDM-2SD38.534.3TrSMIM-2SD45.434.4n=724 (M/F =315/409)18–40L1TrSMAM-2SD110.470.1TrSMDM-2SD36.229.8TrSMIM-2SD34.625.97Benjamin et al. (36)2017IndiaSliceOmatic (−29 to 150 HU)Healthy liver and kidney donorsn=275 (M/F =139/136)18–55L3TrSMAM-2SD95.072.6TrSMIM-2SD36.530.2†, the corresponding cut-off points (M-2SD) are calculated based on the mean and standard deviation values reported in the article. CT, computed tomography; F, female; HU, Hounsfield Unit; M, male; M-2SD, mean minus two standard deviations; P5, 5th percentile; TrSMA, trunk skeletal muscle area (cm2); TrSMD, trunk skeletal muscle density (HU); TrSMI, trunk skeletal muscle index (cm2/m2). Studies establishing CT-derived cut-off points at L3 or L1 level to evaluate trunk muscle metrics among healthy individuals †, the corresponding cut-off points (M-2SD) are calculated based on the mean and standard deviation values reported in the article. CT, computed tomography; F, female; HU, Hounsfield Unit; M, male; M-2SD, mean minus two standard deviations; P5, 5th percentile; TrSMA, trunk skeletal muscle area (cm2); TrSMD, trunk skeletal muscle density (HU); TrSMI, trunk skeletal muscle index (cm2/m2). Our study has several strengths. Firstly, the sample size exceeded 4,000 Chinese adults and encompassed a wide age spectrum. Secondly, the percentile charts were created using the LMS method, offering age- and sex-specific reference values of CT-derived trunk muscle metrics pertinent to the Chinese population aged 25 to 75 years old. Importantly, by adopting multi-stage cluster sampling rather than the retrospective selection of CT scans based on predefined criteria, our approach minimized selection bias. The study also has several limitations, notably its concentration on Chinese participants and a bias towards older demographics, potentially leading to preselection bias and less accurate percentile estimates for younger groups. Future research should include a broader range of younger individuals to refine the accuracy and applicability of the results. This was a cross-sectional study. Therefore, the age-related decreases reported here reflect differences between participants of different ages rather than within-person longitudinal changes. Although the standardized differences per year describe age-related patterns, they should not be interpreted as true annual rates of muscle loss. Longitudinal studies are required to confirm temporal trajectories of myopenia and myosteatosis. Lastly, the M-2SD approach provides reference-based thresholds derived from young adults and has been widely used in CT body composition and sarcopenia research. However, thresholds derived in older adults and anchored to objective measures of muscle strength, physical performance, or adverse clinical outcomes may offer greater clinical relevance. Future studies should further validate CT-derived thresholds in older populations using functional and outcome-based data. Conclusions This study determined age- and sex-specific percentile reference values for trunk muscle area and density at the L1 and L3 levels for a Chinese adult population aged 21–80 years. We found distinct sex-specific age-related patterns in trunk myopenia and myosteatosis, with more pronounced decreases in muscle area and density observed in specific older age intervals. These results offer crucial benchmarks and reference values that could enhance future studies on trunk muscle in myopenia and myosteatosis, as well as help better define trunk muscle sarcopenia. Supplementary The article’s supplementary files as