Abstract
Background
Loss of muscle mass is a known feature of sarcopenia and predicts poor clinical outcomes. While muscle metrics can be derived from routine CT images, sex-specific reference values at multiple vertebral levels over a wide age range are lacking.
Objective
To provide reference values for skeletal muscle mass and attenuation on thoracic and abdominal computed tomography (CT) scans in the community-based Framingham Heart Study (FHS) cohort to aid in the identification of sarcopenia.
Materials and Methods
This secondary analysis of a prospective trial describes muscle metrics by age and sex for participants from the FHS without prior history of cancer who underwent at least one CT scan between 2002 and 2011. Using two previously validated machine learning algorithms followed by human quality assurance, skeletal muscle was analyzed on a single axial CT image per level at the 5th, 8th, 10th thoracic, and 3rd lumbar vertebral body (T5, T8, T10, L3). Cross-sectional muscle area (CSMA, cm2), mean skeletal muscle radio-attenuation (SMRA, Hounsfield Units [HU]), skeletal muscle index (SMI, cm2/m2), and skeletal muscle gauge (SMG, SMRA⋅SMI) were calculated. Measurements were summarized by age group (<45, 45–54, 55–64, 65–74, ≥75 years), sex, and vertebral level. Models enabling the calculation of age-, sex-, and vertebral level specific reference values were created and embedded into an open access online web application.
Results
The cohort consisted of 3,804 participants (1,917 [50.4%] males; mean age 55.6 ± 11.8 years, range 33–92 years) and 7,162 CT scans. Muscle metrics qualitatively decreased with increasing age and female sex.
Conclusion
This study established age- and sex-specific reference values for CT-based muscle metrics at thoracic and lumbar vertebral levels. These values may be used in future research investigating the role of muscle mass and attenuation in health and disease, and to identify sarcopenia.
Keywords: Reference Values, Skeletal muscle, Sarcopenia, Muscular Diseases, Multidetector Computed Tomography, Abdomen, Thorax, Algorithms, Male, Female
Introduction
Sarcopenia is a condition characterized by reduced muscle mass and strength that leads to decreased physical performance1–4 and has been identified as a public health concern.2,5 Sarcopenia and low muscle mass have been linked to decreased survival and lower quality of life in patients with cancer,6,7 as well as increased hospitalization costs in older adults.8
Various definitions for sarcopenia exist, and muscle mass is frequently considered in addition to muscle strength.1–4 While lean mass can be assessed using dual-energy X-ray absorptiometry or MRI,9,10 it is also possible to derive muscle mass and attenuation (also referred to as “density”) metrics from routine axial computed tomography (CT) images.1,11 Reference values are needed to both interpret individual measurements and to define sarcopenia based on thresholds at the mean – 2 standard deviations.12,13,14 Published reference values for muscle mass and attenuation on CT images in U.S. populations included measurements at the lower thoracic and lumbar vertebral levels.12,15,16 However, data on muscle metrics at higher thoracic vertebral levels in U.S. populations are lacking.
The National Heart, Lung, and Blood Institute’s Framingham Heart Study (FHS) is a well-described, longitudinal, community-based cohort study launched in 1948 to evaluate risk factors for cardiovascular disease.17 It follows multiple generations of participants and continues to be an active study. The FHS eligibility criteria and cohort demographics have been described in previous works.18–20 Various reference values have been proposed based on analyses of FHS participants in other fields.21,22
The purpose of this study was to estimate age- and sex-specific reference values for skeletal muscle at the T5, T8, T10, and L3 vertebral levels amongst participants of the FHS without a history of malignancy who underwent CT scans, and to provide an open access online web application to dynamically generate these values based on user-defined inputs.
Materials and Methods
The institutional review board of Massachusetts General Hospital approved this secondary analysis of a prospective trial, which was compliant with the Health Insurance Portability and Accountability Act.
Framingham Heart Study
A list of all previous publications using FHS (ClinicalTrials.gov: NCT00005121) data is available at https://www.framinghamheartstudy.org/fhs-bibliography/. The FHS cohorts who underwent multi-detector CT and provided informed written consent include 4,452 participants from the Offspring, Generation 3, Omni 1, and Omni 2 cohorts who were mobile, weighed less than 352 pounds, and were either men preferably above 35 years of age or non-pregnant women above 40 years of age. Of those, we excluded participants with a history of cancer, except for non-malignant skin tumors (Figure 1). Non-contrast scans were acquired with participants in supine position and a phantom23 under their back between 2002–2005 (CT-1) and 2008–2011 (CT-2) following standardized protocols (see “CT Acquisition Parameters”, Supplemental Methods, Supplemental Digital Content 1, for a detailed description of acquisition protocols).24,25
Figure 1.
Flow diagram of inclusion and exclusion criteria. 13,184 scans of 4,452 participants were initially considered for further analysis. * The most recent CT scan suitable for analysis of muscle was chosen at each vertebral level. Abbreviations: FHS – Framingham Heart Study; CT – Computed Tomography; QA – Quality Assurance.
Data Collection and Definitions
Image Analysis
To avoid misclassification of the phantom as muscle, we first removed the phantoms from the images using an object detection algorithm (https://github.com/FintelmannLabDevelopmentTeam/PhantomDeletionFHS [commit c195272]). We then applied two previously validated fully automated deep-learning pipelines26,27 to quantify and characterize skeletal muscle on images of the thorax with a slice thickness of 2.5 mm and images of the abdomen with a slice thickness of 5 mm. Briefly, each algorithm consists of two steps. First, a network selects an axial image at the center of the T5, T8, T10, or L3 vertebral body. Second, another network segments skeletal muscle on the selected axial images.26,27
Quality Assurance
Two analysts (A.S.W.D., P.E.T.), trained by a board-certified radiologist (F.J.F.) with seven years of experience, independently reviewed all segmentations. They identified the most recent segmentation for each participant and validated it by comparing the label map with the corresponding CT image (Figure 2). If necessary, the analysts could manually edit the level selection and segmentation proposed by the algorithm. Manual corrections of the muscle segmentation were performed with a threshold of -29 to 150 Hounsfield Units (HU). If the initial segmentation was not salvageable by manual editing, the analysts identified another scan of the participant and repeated the validation procedure. The analysts excluded participants if no scan was salvageable by manual editing and logged reasons for exclusion (see “Reasons for Exclusion during Quality Assurance”, Supplemental Methods, Supplemental Digital Content 1, where the individual exclusion criteria are summarized). With a minimum 60-day wash-out period, both analysts each re-evaluated 500 randomly selected segmentations (125 per vertebral level).
Figure 2.
Illustration of image analysis and quality assurance.
Muscle Metrics
We calculated the cross-sectional muscle area (CSMA [cm2]) and the mean skeletal muscle radio-attenuation (SMRA [HU]) at vertebral levels T5, T8, T10, and L3. We defined the skeletal muscle index (SMI [cm2/m2]) as CSMA divided by the participant’s squared height in meters and the skeletal muscle gauge (SMG [SMRA⋅SMI]).28
Definition of Participant Characteristics
Participant characteristics were obtained from the FHS documentation closest to the date of the CT scans (see “Cohort Data Collection”, Supplemental Methods, Supplemental Digital Content 1, where a detailed description of how individual characteristics were extracted is given).
Statistical Analysis
All analyses were performed by P.E.T. under the supervision of N.D.M. using R version 4.1.2 (The R Foundation for Statistical Computing, Vienna, Austria). The analysis code is publicly available (https://github.com/FintelmannLabDevelopmentTeam/MuscleReferenceValuesFHS [commit 70fd327]).
We assessed inter- and intra-reader agreement between the two analysts who performed the quality assurance by calculating Cohen’s kappa and the respective 95% confidence interval (95% CI). 29 Intraclass correlation coefficients (ICC) were used to assess agreement for cross-sectional muscle area (CSMA).
We calculated descriptive statistics for CSMA, SMI, SMRA, and SMG, stratifying by age groups (<45, 45–54, 65–74, and ≥75 years), sex, and vertebral level. We repeated this descriptive analysis in pre-specified subgroup analyses to provide reference values for racial/ethnic groups and specific FHS cohorts.
In line with recommendations of the WHO Growth Curve Publication Group from 2006, we used the LMSP method from the open source GAMLSS R package (version 5.4.3) to estimate the relationship between age and muscle metrics.30–32 For these models, we considered only participants between age 38 and 80 to prevent edge effects.32 Model parameters were estimated automatically by minimizing the generalized Akaike information criterion (GAIC), as suggested by Rigby and Stasinopoulos.33 We performed a grid search to find a suitable smoothing penalty for the GAIC and selected models based on the Schwarz Bayesian Information Criterion, the Akaike Information Criterion, and visual inspection of resulting centile curves, worm plots, and Q-statistics34,35 (see “Reference Curves”, Supplemental Methods, where the model selection procedure is described in more detail). We used the models to estimate age- and sex-specific reference centile curves at the 3rd, 15th, 50th, 85th, and 97th percentile for CSMA, SMI, SMRA, and SMG at the T5, T8, T10, and L3 vertebral levels.
All models were incorporated into an open access web application for dynamic centile estimation as a function of sex, age, vertebral level, and muscle metric using R Shiny version 1.7.2.
Results
Participant Characteristics
Following quality assurance, a total of 12,505 validated segmentations from 7,162 scans in 3,804 participants were included (Figure 1). The study cohort included 1,917 men (50.4%) and 1,887 women (49.6%) (Table 1 and “Supplemental Table 3“, Supplemental Digital Content for a detailed breakdown of comorbidities). Mean age was 54 ± 11.7 (standard deviation [SD]) years in women and 57.3 ± 11.6. years in men. Most participants were white (1,740 men [91%]; 1,686 women [89%]) and were part of the Generation 3 cohort (1,181 men [62%]; 967 women [51%]).
Table 1:
Clinical Characteristics of Framingham Heart Study participants with muscle measurements derived from axial CT images
| Characteristic | Male, N=1,917 | Female, N=1,887 |
|---|---|---|
| Age | 54.0 (11.7) | 57.3 (11.6) |
| Height (m) | 1.70 (0.07) | 1.56 (0.06) |
| Weight (kg) | 90 (16) | 72 (16) |
| BMI | 31.2 (5.0) | 29.7 (6.4) |
| Waist circumference (in) | 31 (18) | 30 (17) |
| Current Smoker | 175 (9.4) | 156 (8.6) |
| Drinking Alcohol | 1,462 (77) | 1,260 (67) |
| Hypertension | 855 (45) | 763 (41) |
| Postmenopausal | 0 (0) | 1,214 (67) |
| Diabetes | 247 (13) | 212 (11) |
| Metabolic syndrome | 395 (21) | 320 (17) |
| Framingham Risk Score | 0.13 (0.11) | 0.07 (0.07) |
| Physical Activity Index | 37 (8) | 35 (5) |
| Race and Ethnicity | ||
| American Indian or Alaskan Native | 3 (0.2) | 0 (0) |
| Asian | 31 (1.6) | 34 (1.8) |
| Black or African American | 25 (1.3) | 41 (2.2) |
| Hispanic or Latino | 35 (1.8) | 45 (2.4) |
| Native Hawaiian or Other Pacific Islander | 1 (<0.1) | 0 (0) |
| Other / Not reported | 82 (4.3) | 81 (4.3) |
| White | 1,740 (91) | 1,686 (89) |
| FHS Cohort | ||
| Offspring (Generation 2) | 629 (33) | 770 (41) |
| Generation 3 | 1,181 (62) | 967 (51) |
| Omni 1 | 44 (2.3) | 71 (3.8) |
| Omni 2 | 63 (3.3) | 79 (4.2) |
Sex-specific cohort characteristics of participants with muscle measurements for at least one vertebral level (n=3,804). If measurements at different vertebral levels were derived from scans taken at different time points, the data acquired at the most recent time point is presented here.
Categorical variables are summarized as n (%). Continuous variables are described with mean (standard deviation). Abbreviations: BMI – body mass index; FHS – Framingham Heart Study. Additional characteristics for participants with measurements available at the T5, T8, T10, and L3 vertebral levels and a description of how much data was missing are given in the “Supplemental Table 3“, Supplemental Digital Content. Race was self-reported. “Other” encompasses participants self-reporting as other than the given racial and ethnic groups and participants self-reporting to belong to more than one racial or ethnic group.
Quality Assurance
The main reason for the exclusion of measurements from a CT scan at a given vertebral level during quality assurance was that the vertebral level of interest was not included in the field of view. This limited z-axis coverage affected 56.5% of all scans at T5 and <0.5% of the other vertebral levels (see “Reasons for Exclusion”, Supplemental Digital Content 1 for a summary of reasons for exclusion). The second most common reason for exclusion was the incomplete visualization of muscle at the vertebral level of interest which affected 28% of scans at the T8 level, 17% at the T10 level, 1% of scans at the T5 level, and <1% of scans at the L3 level. Overall, muscle could be measured on at least one CT scan for 59.3% of participants at the T5 vertebral level and for over 81% of participants at the other vertebral levels (Figure 1).
ICCs were 0.997 (95% CI: 0.995, 0.997) for both analysts’ intrareader agreement of CSMA measurements. The ICC for interreader agreement was 0.997 (95% CI 0.997, 0.998). Intrareader agreement on the decision to exclude images was k=0.9, p<0.001 for one analyst and k=1, p<0.001 for the other. Interreader agreement on scan exclusion was k=0.93, p<0.001.
Empirical Reference Values
The mean, SD, median, and interquartile range (IQR) of CSMA, SMI, SMRA, and SMG are listed in Table 2 for women and Table 3 for men, stratified by age group and vertebral level. The corresponding violin plots for CSMA and SMI are shown in Figure 3, and for SMRA and SMG in Figure 4. All metrics of muscle qualitatively decreased with age at all vertebral levels. Measurements of CSMA, SMI, and SMG tended to decrease when moving downwards along the spine from T5 to T10, with another increase toward L3. Mean CSMA ranged from 54.9 cm2 (women, ≥75, at T10) to 251.2 cm2 (men, <45, at T5). Mean SMI ranged from 23.9 cm2/m2 (women, ≥75, at T10) to 86. 9 cm2/m2 (men, <45, at T5). Mean SMRA ranged from 23.2 HU (women, ≥75, at T10) to 46 HU (men, <45, at T5). Mean SMG ranged from 561.7 HU⋅cm2/m2 (women, ≥75, at T10) to 4,017.5 HU⋅cm2/m2 (men, <45, at T5).
Table 2:
Empirical reference values for skeletal muscle metrics in female participants of the Framingham Heart Study
| T5, N = 1,153 | T8, N = 1,811 | T10, N = 1,837 | L3, N = 1,859 | |||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Age (Years) | <45 | 45–54 | 55–64 | 65–74 | ≥75 | <45 | 45–54 | 55–64 | 65–74 | ≥75 | <45 | 45–54 | 55–64 | 65–74 | ≥75 | <45 | 45–54 | 55–64 | 65–74 | ≥75 |
| N | 88 | 360 | 355 | 237 | 113 | 265 | 569 | 489 | 319 | 169 | 277 | 583 | 495 | 315 | 167 | 279 | 588 | 500 | 321 | 171 |
| CSMA (cm2) | ||||||||||||||||||||
| Mean (SD) | 156.7 (22.2) | 145.0 (19.8) | 134.8 (17.2) | 128.6 (15.7) | 119.2 (16.7) | 83.7 (12.4) | 79.3 (12.5) | 72.4 (10.9) | 68.1 (10.4) | 64.2 (10.7) | 70.9 (11.3) | 67.5 (11.2) | 62.4 (9.9) | 58.6 (9.7) | 54.9 (9.2) | 120.9 (18.4) | 119.5 (18.5) | 110.6 (16.3) | 104.5 (15.8) | 99.5 (16.0) |
| Median (IQR) | 155.7 (23.5) | 144.4 (24.2) | 134.4 (22.7) | 128.5 (21.6) | 118.2 (17.2) | 83.7 (15.0) | 78.7 (16.9) | 71.7 (14.0) | 67.2 (15.4) | 62.7 (13.8) | 70.0 (14.5) | 66.9 (15.6) | 61.8 (13.5) | 58.6 (13.0) | 54.9 (11.1) | 118.5 (22.1) | 117.6 (24.5) | 109.1 (20.2) | 103.6 (20.5) | 97.4 (18.5) |
| SMI (cm2/m2) | ||||||||||||||||||||
| Mean (SD) | 63.4 (9.3) | 58.1 (8.2) | 55.8 (7.9) | 54.7 (7.3) | 52.0 (7.4) | 33.5 (5.2) | 31.9 (5.2) | 30.0 (5.0) | 28.9 (4.8) | 28.0 (4.6) | 28.4 (4.8) | 27.2 (4.6) | 25.9 (4.4) | 24.8 (4.2) | 23.9 (4.0) | 48.2 (7.1) | 48.0 (7.2) | 45.8 (6.8) | 44.3 (6.5) | 43.5 (6.7) |
| Median (IQR) | 62.2 (11.2) | 57.6 (10.1) | 55.3 (9.5) | 54.3 (10.2) | 51.7 (9.4) | 33.2 (6.3) | 31.4 (6.6) | 29.6 (6.3) | 28.3 (6.9) | 27.7 (6.4) | 27.5 (5.5) | 26.6 (6.2) | 25.5 (6.0) | 24.6 (5.5) | 23.8 (5.6) | 46.9 (8.4) | 47.3 (10.1) | 45.2 (8.4) | 43.6 (8.2) | 42.3 (7.5) |
| SMRA (HU) | ||||||||||||||||||||
| Mean (SD) | 42.3 (5.1) | 40.3 (5.0) | 38.6 (4.9) | 37.5 (4.7) | 35.2 (4.8) | 36.3 (6.2) | 31.9 (6.2) | 28.0 (5.8) | 26.1 (5.4) | 24.0 (5.7) | 37.0 (5.9) | 32.1 (6.6) | 28.2 (6.2) | 25.8 (5.5) | 23.2 (6.1) | 42.6 (5.5) | 38.1 (7.2) | 33.4 (6.8) | 29.3 (6.6) | 25.2 (6.8) |
| Median (IQR) | 42.7 (5.8) | 40.1 (6.1) | 38.5 (5.8) | 37.2 (5.3) | 34.8 (4.8) | 36.6 (7.7) | 31.7 (8.9) | 27.9 (8.0) | 26.2 (6.8) | 24.1 (6.9) | 36.8 (8.7) | 31.7 (8.6) | 27.8 (8.0) | 25.7 (6.6) | 22.9 (7.8) | 43.4 (6.6) | 38.9 (9.7) | 33.6 (9.5) | 29.2 (8.9) | 25.3 (9.8) |
| SMG (HU⋅cm2/m2) | ||||||||||||||||||||
| Mean (SD) | 2,693.9 (575.1) | 2,352.0 (486.3) | 2,163.0 (480.4) | 2,055.1 (415.8) | 1,842.3 (436.3) | 1,224.4 (308.8) | 1,026.9 (290.0) | 851.2 (266.3) | 763.7 (243.7) | 679.2 (224.9) | 1,052.7 (251.1) | 877.3 (258.4) | 736.5 (230.6) | 645.8 (199.8) | 561.7 (193.5) | 2,047.9 (376.3) | 1,824.0 (416.2) | 1,524.5 (367.9) | 1,297.7 (346.7) | 1,111.6 (374.5) |
| Median (IQR) | 2,643.4 (586.1) | 2,365.5 (609.1) | 2,107.6 (586.2) | 2,010.7 (529.5) | 1,817.4 (490.7) | 1,233.9 (431.1) | 1,021.5 (406.8) | 808.2 (349.2) | 730.0 (316.5) | 664.5 (332.5) | 1,031.3 (347.7) | 857.3 (313.0) | 703.4 (293.4) | 613.7 (234.9) | 537.1 (260.7) | 2,026.2 (495.0) | 1,828.6 (494.7) | 1,509.5 (485.9) | 1,306.3 (499.7) | 1,043.6 (489.6) |
Reference values for skeletal muscle metrics in female participants of the Framingham Heart Study at the T5, T8, T10, and L3 vertebral levels. The number of participants differs between vertebral levels due to exclusion during quality assurance.
Abbreviations: CSMA – cross-sectional muscle area; IQR – interquartile range; HU – Hounsfield Units; SD – standard deviation; SMG – skeletal muscle gauge; SMI – skeletal muscle index; SMRA – skeletal muscle radio-attenuation.
Table 3:
Empirical reference values for skeletal muscle metrics in male participants of the Framingham Heart Study
| T5, N = 1,122 | T8, N = 1,317 | T10, N = 1,513 | L3, N = 1,893 | |||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Age (Years) | <45 | 45–54 | 55–64 | 65–74 | ≥75 | <45 | 45–54 | 55–64 | 65–74 | ≥75 | <45 | 45–54 | 55–64 | 65–74 | ≥75 | <45 | 45–54 | 55–64 | 65–74 | ≥75 |
| N | 206 | 395 | 280 | 176 | 65 | 238 | 455 | 325 | 217 | 82 | 327 | 530 | 351 | 219 | 86 | 471 | 639 | 419 | 262 | 102 |
| CSMA (cm2) | ||||||||||||||||||||
| Mean (SD) | 251.2 (34.1) | 233.9 (28.1) | 213.6 (26.5) | 194.7 (27.1) | 177.9 (29.5) | 141.7 (22.6) | 133.6 (20.1) | 120.7 (17.7) | 108.4 (18.3) | 101.8 (15.4) | 114.6 (17.0) | 108.8 (16.4) | 100.8 (15.2) | 89.3 (14.8) | 81.8 (15.6) | 194.3 (26.0) | 188.6 (25.8) | 180.7 (24.5) | 163.4 (21.5) | 153.4 (24.8) |
| Median (IQR) | 249.6 (44.8) | 231.2 (33.9) | 212.8 (37.8) | 191.4 (36.3) | 177.6 (36.5) | 140.1 (30.2) | 131.5 (25.0) | 120.2 (23.9) | 106.8 (21.6) | 102.3 (17.7) | 114.5 (24.1) | 107.4 (20.2) | 100.7 (18.7) | 88.8 (21.4) | 80.7 (22.0) | 193.5 (35.1) | 186.5 (32.2) | 178.8 (31.3) | 163.4 (27.0) | 153.6 (33.8) |
| SMI (cm2/m2) | ||||||||||||||||||||
| Mean (SD) | 86.9 (12.2) | 80.9 (10.7) | 74.5 (9.5) | 69.7 (9.8) | 64.7 (10.0) | 49.1 (8.3) | 46.3 (7.6) | 42.1 (6.6) | 38.9 (6.9) | 37.4 (5.8) | 39.5 (6.4) | 37.6 (6.0) | 35.2 (5.7) | 32.1 (5.5) | 30.1 (6.0) | 66.3 (9.0) | 64.8 (8.9) | 62.9 (8.6) | 58.7 (7.8) | 56.0 (8.5) |
| Median (IQR) | 87.5 (16.2) | 80.4 (13.8) | 74.6 (12.7) | 68.7 (13.0) | 63.2 (12.8) | 48.2 (11.8) | 45.6 (10.3) | 41.6 (8.3) | 38.5 (9.0) | 37.0 (7.1) | 39.7 (8.6) | 36.8 (8.4) | 35.0 (7.2) | 31.8 (7.1) | 28.2 (8.9) | 65.7 (12.8) | 64.1 (10.9) | 61.8 (10.2) | 58.6 (11.0) | 56.0 (10.9) |
| SMRA (HU) | ||||||||||||||||||||
| Mean (SD) | 46.0 (5.6) | 44.5 (4.7) | 42.5 (4.9) | 41.3 (5.2) | 38.3 (4.8) | 38.8 (5.9) | 36.4 (5.7) | 33.8 (5.5) | 31.6 (5.8) | 28.9 (4.9) | 39.6 (6.1) | 36.6 (6.1) | 33.3 (6.0) | 30.3 (6.1) | 27.4 (6.4) | 45.2 (5.6) | 42.0 (5.9) | 38.7 (6.2) | 33.9 (6.6) | 30.6 (6.7) |
| Median (IQR) | 45.7 (4.2) | 44.2 (4.3) | 42.5 (5.1) | 41.1 (5.6) | 38.6 (5.8) | 38.5 (5.6) | 36.2 (6.2) | 33.9 (6.6) | 31.9 (7.2) | 28.6 (6.4) | 39.0 (7.7) | 36.3 (7.6) | 33.2 (7.2) | 29.5 (7.5) | 27.3 (7.0) | 46.0 (7.1) | 42.6 (8.2) | 39.6 (8.5) | 34.5 (8.9) | 30.5 (10.2) |
| SMG (HU⋅cm2/m2) | ||||||||||||||||||||
| Mean (SD) | 4,017.5 (881.8) | 3,611.5 (685.6) | 3,185.2 (650.8) | 2,892.4 (627.2) | 2,490.2 (548.8) | 1,921.3 (521.3) | 1,702.9 (458.9) | 1,437.8 (387.2) | 1,251.2 (400.8) | 1,093.6 (300.0) | 1,574.1 (387.4) | 1,388.5 (377.5) | 1,188.1 (351.5) | 988.8 (309.0) | 846.1 (320.8) | 3,000.0 (560.3) | 2,725.4 (553.9) | 2,437.8 (550.3) | 2,004.8 (520.3) | 1,719.5 (504.0) |
| Median (IQR) | 3,976.1 (973.6) | 3,552.4 (784.0) | 3,178.5 (728.6) | 2,855.4 (780.5) | 2,475.1 (742.8) | 1,891.7 (663.2) | 1,638.7 (549.3) | 1,412.8 (481.8) | 1,230.6 (468.5) | 1,052.7 (410.4) | 1,566.7 (483.2) | 1,352.3 (466.0) | 1,173.0 (431.8) | 963.0 (437.3) | 794.6 (351.1) | 3,001.8 (692.1) | 2,728.5 (672.4) | 2,448.2 (705.6) | 2,017.6 (712.9) | 1,705.0 (633.0) |
Reference values for skeletal muscle metrics in male participants of the Framingham Heart Study at the T5, T8, T10, and L3 vertebral levels. The number of participants differs between vertebral levels due to exclusion during quality assurance.
Abbreviations: CSMA – cross-sectional muscle area; IQR – interquartile range; HU – Hounsfield Units; SD – standard deviation; SMG – skeletal muscle gauge; SMI – skeletal muscle index; SMRA – skeletal muscle radio-attenuation.
Figure 3.
Violin plots visualizing the distribution of A) cross-sectional muscle area and B) skeletal muscle index in participants of the Framingham Heart Study. The black horizontal lines indicate the 3rd, 25th, 50th (median) and 75th percentile, respectively. The black centered dot marks the mean. The red horizontal line shows the mean – 2 standard deviation thresholds.
Figure 4.
Violin plots visualizing the distribution of A) skeletal muscle radio-attenuation and B) skeletal muscle gauge in participants of the Framingham Heart Study. The black horizontal lines indicate the 3rd, 25th, 50th (median) and 75th percentile, respectively. The black centered dot marks the mean. The red horizontal line shows the mean – 2 standard deviation threshold.
Numerous cohort summary and reference value tables, as well as violin plots for the following subgroups are presented in “Supplement SA: Subgroup Analyses”, Supplemental Digital Content 1: Offspring cohort, Generation 3 cohort, Asian, Black or African American, and White participants.
Age- and Sex-specific Reference Centile Curves
Reference centile curves for CSMA and SMI are visualized in Figure 5. Reference centile curves for SMRA and SMG are visualized in Figure 6. Measurements of CSMA, SMI, SMRA, and SMG of patients with known age and sex may be compared against distributions generated by the LMSP models to calculate how the measurement relates to the median (z-score). The models are embedded into an R Shiny App under https://muscle-metrics.mgh.harvard.edu/ (see https://github.com/FintelmannLabDevelopmentTeam/MuscleReferenceValuesFHS_ShinyApp [commit 6773803] for running the app offline) to allow the calculation of z-scores via an open access web application.
Figure 5.
Age- and sex-specific reference centile curves for A) cross-sectional muscle area (CSMA) and B) skeletal muscle index (SMI) in participants of the Framingham Heart Study aged 38 to 80 years at the T5, T8, T10, and L3 vertebral level. The curves for CSMA and SMI appear to decrease with age in all subgroups.
Figure 6.
Age- and sex-specific reference centile curves for A) skeletal muscle radio-attenuation (SMRA) and B) skeletal muscle gauge (SMG) in participants of the Framingham Heart Study aged 38 to 80 years at the T5, T8, T10, and L3 vertebral level. The curves for SMRA and SMG appear to decrease with age in all subgroups.
Interpreting reference values
Table 4 shows hypothetical measurements of SMI at the T8 vertebral level on serial CT scans acquired for the purpose of lung cancer screening in a female (patient #1) and a male (patient #2) at ages 50, 60, 70, and 80 years. The measurements were entered into the online tool and z-scores were calculated. Both patients show a declining SMI with increasing age (Patient #1: 32 cm2/m2 at 50 years of age to 19 cm2/m2 at 80 years of age; Patient #2: 49.6 cm2/m2 at 50 years of age to 38.3 cm2/m2 at 80 years of age). Although SMI measurements in patient #2 declined with age, his z-scores stayed stable at around 0.52, indicating an above average thoracic muscle mass compared to participants of the FHS of same age and sex at all time points. However, patient #1 showed a decline in both SMI and the z-scores, indicating that intra-individual loss of thoracic muscle mass was happening at a higher-than-average rate compared to that of female participants of the Framingham Heart Study. Additionally, at age 75, the individual z-score falls below -2, indicating the presence of sarcopenia.12,13
Table 4:
Reference Values Example
| Patient #1: Female | Patient #2: Male | |||
|---|---|---|---|---|
| Age | SMI at T8 | Z-Score | SMI at T8 | Z-Score |
| 50 | 32 cm2/m2 | 0.13 | 49.6 cm2/m2 | 0.52 |
| 60 | 30 cm2/m2 | 0.08 | 45.5 cm2/m2 | 0.53 |
| 70 | 26 cm2/m2 | −0.5 | 41.7 cm2/m2 | 0.52 |
| 80 | 19 cm2/m2 | −2.03 | 38.3 cm2/m2 | 0.52 |
Hypothetical serial measurements of the skeletal muscle index (SMI) at the level of the eighth thoracic vertebral body in two patients undergoing a low dose CT scan for routine lung cancer screening at four time points. Z-Scores were calculated using the free online tool available at *link, blinded for peer review*. Z-Scores <-2 are highlighted in red to indicate potential presence of sarcopenia. Abbreviations: SMI – skeletal muscle index; T8 – the level of the eighth thoracic vertebral body
Discussion
This study analyzed skeletal muscle mass and attenuation at the level of the 5th, 8th, 10th thoracic, and the 3rd lumbar vertebral body on a total of 7,162 scans obtained from 3,804 participants of the Framingham Heart Study. The participants selected for analysis had no prior history of cancer and ranged in age from 33 to 92 years. An open access online web application allows for the calculation of age- and sex-specific reference values for muscle metrics at multiple thoracic and lumbar vertebral levels in patients from 38 to 80 years to use in future research.
Reference values for muscle mass on CT of U.S.-based populations have been previously described.12,15,16 Our study adds to the existing literature in several ways. First, while previous work focused on muscle mass at lumbar12,15,16 and lower thoracic12 vertebral levels, this study also analyzed the T5 and T8 level. Second, participants in the Framingham Heart Study have been described in detail (see “Supplemental Table 3“, Supplemental Digital Content 1 for a detailed breakdown of comorbidities), which facilitates the interpretation of reference values derived from this cohort. Third, due to the prospective nature of the study, imaging protocols were standardized, eliminating the presence of potential confounders on measurements.36 Fourth, every segmentation generated by the previously validated and published deep-learning algorithms was reviewed and edited as necessary with excellent inter- and intrareader agreement to ensure high quality measurements. Lastly, an open access online web application facilitates future research by allowing for comparison of muscle measurements in patients aged 38 to 80 with age-, sex-, and vertebral level-specific reference values via z-scores and centile estimates.
In line with other studies based on cohorts of similar size and age distribution, our study documents an influence of age and sex on muscle mass and attenuation.15,37 Furthermore, the mean of CSMA and SMI at the L3 level in participants <45 years of age was comparable to other cohorts.12,16 Larger differences were noted when comparing the mean CSMA and SMI at the T10 level to a smaller and younger cohort which used a different measurement technique, limiting comparability.12 Additionally, the mean SMRA measurements match the range described in other cohorts.12,16
Multi-level assessment at T5, T8, T10, and L3 and the ability to rank muscle measurements based on reference values provide additional benefits compared to measurement on L3 alone. First, the aggregation of measurements from multiple vertebral levels can increase their predictive value in clinical research.28 Second, while population-wide opportunistic screening for sarcopenia and low muscle on existing routine CT data have previously been discussed,2,14,16,38,39 reference values on thoracic levels are needed to avoid limiting such screening to only abdominal scans, potentially missing parts of the vulnerable target population receiving yearly routine lung cancer screening40,41 or even a single CT scan for diagnostic purposes.
Some sarcopenia definitions recommend that reference values stem from healthy and young (18–40 years) individuals.13 Although preselected and arguably healthy, most of this study’s participants do not fall in this age range. However, reference values from both young and same-age populations are used to assess bone mineral density via t- and z-scores, respectively, each with their own utility.42 CT measurements of CSMA and SMI may be used as surrogate parameters for muscle mass and z-scores of -2 could be used as a cut-off for the diagnosis of sarcopenia.11–13
In addition, CT-based measurement of SMRA has been used to estimate fat infiltration in the muscle (myosteatosis),43 while SMG has been proposed as a composite metric that combines the advantages of SMI and SMRA.28 As Table 4 illustrates, longitudinal measurements can deliver insights into the development of individual metrics of muscle mass and attenuation compared to the respective group of FHS participants of same sage and sex over time and show where early intervention is needed to counter low muscle mass and potential myosteatosis. Future studies should evaluate the potential utility of our age- and sex-specific reference values for CSMA, SMI, SMRA, and SMG in combination with metrics of physical performance regarding the role of muscle mass and attenuation in health and disease.
Our study results need to be interpreted in the context of existing limitations. First, our cohort consists primarily of white participants. Race is known to influence muscle mass15 and further research should be conducted in more diverse cohorts to provide reference values from other subgroups. Second, the presence of first-degree relatives in our cohort further limits cohort diversity. Third, CT scans were taken almost two decades ago, and birth cohort effects might apply. Fourth, CT-1 scans focused on the heart, and 56.5% of scans became unavailable for measuring muscle at the T5 vertebral level. Finally, it is important to note that the validated segmentation tools employed in this study were designed to quantify muscle on an axial image at a vertebral level as opposed to volumetric analysis.
In summary, this study provides “normative” data of skeletal muscle mass and attenuation at the level of the 5th, 8th, 10th thoracic, and the 3rd lumbar vertebral body derived from a large cohort of participants without cancer for potential use in evolving definitions of sarcopenia. Clearly defined measurement technique and free online calculation tools providing age- and sex-adjusted reference values pave the way for future opportunistic screening for low muscle mass on both abdominal and thoracic CT scans.
Supplementary Material
Sources of Funding:
The Framingham Heart Study is supported by Contract No. HHSN268201500001I from the National Heart, Lung, and Blood Institute (NHLBI) with additional support from other sources. This work was supported by the FHS Core Contract (NHLBI award# 75N92019D00031) and National Institute of Health (NIH) grants R01-AR073019 (M.L.B.), R01 AG041658 (E.J.S.), and R01 AR041398 (D.P.K.) from the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS). This manuscript was not prepared in collaboration with investigators of the Framingham Heart Study and does not necessarily reflect the opinions or conclusions of the Framingham Heart Study or the NHLBI.
Footnotes
Conflicts of interest: D.P.K. serves on scientific advisory boards for Reneo, Solarea Bio, and Pfizer, serves on a data monitoring committee for Agnovos, and receives royalties for publication from Wolters Kluwer for UpToDate. F.J.F. receives salary support from the William M. Wood Foundation for unrelated research, consulting fees from Pfizer, and has a related patent. The other authors have no relevant disclosures.
References
- 1.Cruz-Jentoft AJ, Bahat G, Bauer J et al. Writing Group for the European Working Group on Sarcopenia in Older People 2 (EWGSOP2), and the Extended Group for EWGSOP2. Sarcopenia: revised European consensus on definition and diagnosis. Age Ageing 2019;48(1):16–31. Erratum in: Age Ageing. 2019 Jul 1;48(4):601. doi: 10.1093/ageing/afy169. Published January 1, 2004. Accessed March 7, 2022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Chen LK, Liu LK, Woo J, et al. Sarcopenia in Asia: consensus report of the Asian Working Group for Sarcopenia. J Am Med Dir Assoc 2014. Feb;15(2):95–101. doi: 10.1016/j.jamda.2013.11.025. [DOI] [PubMed] [Google Scholar]
- 3.Morley JE, Abbatecola AM, Argiles JM, et al. Society on Sarcopenia, Cachexia and Wasting Disorders Trialist Workshop. Sarcopenia with limited mobility: an international consensus. J Am Med Dir Assoc 2011. Jul;12(6):403–9. doi: 10.1016/j.jamda.2011.04.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Studenski SA, Peters KW, Alley DE, et al. The FNIH sarcopenia project: rationale, study description, conference recommendations, and final estimates. J Gerontol A Biol Sci Med Sci 2014. May;69(5):547–58. doi: 10.1093/gerona/glu010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Beaudart C, Rizzoli R, Bruyère O, et al. Sarcopenia: burden and challenges for public health. Arch Public Health 2014. Dec 18;72(1):45. doi: 10.1186/2049-3258-72-45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Nipp RD, Fuchs G, El-Jawahri A, et al. Sarcopenia Is Associated with Quality of Life and Depression in Patients with Advanced Cancer. Oncologist 2018. Jan;23(1):97–104. doi: 10.1634/theoncologist.2017-0255. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Troschel FM, Jin Q, Eichhorn F, et al. Sarcopenia on preoperative chest computed tomography predicts cancer-specific and all-cause mortality following pneumonectomy for lung cancer: A multicenter analysis. Cancer Med 2021. Oct;10(19):6677–6686. doi: 10.1002/cam4.4207. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Antunes AC, Araújo DA, Veríssimo MT, et al. Sarcopenia and hospitalisation costs in older adults: a cross-sectional study. Nutr Diet 2017. Feb;74(1):46–50. doi: 10.1111/1747-0080.12287. [DOI] [PubMed] [Google Scholar]
- 9.Albano D, Messina C, Vitale J, Sconfienza LM. Imaging of sarcopenia: old evidence and new insights. Eur Radiol 2020 Apr;30(4):2199–2208. Published April 2020. Accessed June 17, 2023. doi: 10.1007/s00330-019-06573-2. [DOI] [PubMed] [Google Scholar]
- 10.Huber FA, Chaitanya K, Gross N, et al. Whole-body composition profiling using a deep learning algorithm: influence of different acquisition parameters on algorithm performance and robustness. Investigative Radiology 2022 Jan; 57(1), 33–43. Published January 2022. Accessed June 17, 2023. doi: 10.1097/RLI.0000000000000799. [DOI] [PubMed] [Google Scholar]
- 11.Zopfs D, Theurich S, Große Hokamp N et al. Single-slice CT measurements allow for accurate assessment of sarcopenia and body composition. Eur Radiol 30, 1701–1708 (2020). Published November 27, 2019. Accessed June 17, 2023. doi: 10.1007/s00330-019-06526-9. [DOI] [PubMed] [Google Scholar]
- 12.Derstine BA, Holcombe SA, Ross BE, et al. Skeletal muscle cutoff values for sarcopenia diagnosis using T10 to L5 measurements in a healthy US population. Sci Rep 2018;8(1):11369. Published July 27, 2018. Accessed March 24, 2022. doi: 10.1038/s41598-018-29825-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Baumgartner RN, Koehler KM, Gallagher D, et al. Epidemiology of sarcopenia among the elderly in New Mexico. Am J Epidemiol 1998. Apr 15;147(8):755–63. Erratum in: Am J Epidemiol 1999 Jun 15;149(12):1161. doi: 10.1093/oxfordjournals.aje.a009520 [DOI] [PubMed] [Google Scholar]
- 14.Fintelmann FJ. Body Composition Analysis on Chest CT Scans: A Value Proposition for Lung Cancer Care. Radiology 2023. Jul;308(1):e231205. doi: 10.1148/radiol.231205. [DOI] [PubMed] [Google Scholar]
- 15.Magudia K, Bridge CP, Bay CP, et al. Population-Scale CT-based Body Composition Analysis of a Large Outpatient Population Using Deep Learning to Derive Age-, Sex-, and Race-specific Reference Curves. Radiology 2021. Feb;298(2):319–329. doi: 10.1148/radiol.2020201640 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Graffy PM, Liu J, Pickhardt PJ, et al. Deep learning-based muscle segmentation and quantification at abdominal CT: application to a longitudinal adult screening cohort for sarcopenia assessment. Br J Radiol 2019. Aug;92(1100):20190327. doi: 10.1259/bjr.20190327. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Dawber TR, Meadors GF, Moore FE. Epidemiological Approaches to Heart Disease: The Framingham Study. Am J Public Health Nations Health 1951. Mar;41(3):279–81. doi: 10.2105/ajph.41.3.279 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Splansky GL, Corey D, Yang Q, et al. The Third Generation Cohort of the National Heart, Lung, and Blood Institute’s Framingham Heart Study: design, recruitment, and initial examination. Am J Epidemiol 2007. Jun 1;165(11):1328–35. doi: 10.1093/aje/kwm021 [DOI] [PubMed] [Google Scholar]
- 19.Kannel WB, Feinleib M, McNamara PM, et al. An investigation of coronary heart disease in families. The Framingham offspring study. Am J Epidemiol 1979. Sep;110(3):281–90. doi: 10.1093/oxfordjournals.aje.a112813 [DOI] [PubMed] [Google Scholar]
- 20.Tsao CW, Vasan RS. Cohort Profile: The Framingham Heart Study (FHS): overview of milestones in cardiovascular epidemiology. Int J Epidemiol 2015. Dec;44(6):1800–13. doi: 10.1093/ije/dyv337 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Vasan RS, Larson MG, Benjamin EJ, et al. Echocardiographic reference values for aortic root size: the Framingham Heart Study. J Am Soc Echocardiogr 1995. Nov-Dec;8(6):793–800. doi: 10.1016/s0894-7317(05)80003-3 [DOI] [PubMed] [Google Scholar]
- 22.Truong QA, Massaro JM, Rogers IS, et al. Reference values for normal pulmonary artery dimensions by noncontrast cardiac computed tomography: the Framingham Heart Study. Circ Cardiovasc Imaging 2012. Jan;5(1):147–54. doi: 10.1161/CIRCIMAGING.111.968610 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Speliotes EK, Massaro JM, Hoffmann U, et al. Liver fat is reproducibly measured using computed tomography in the Framingham Heart Study. J Gastroenterol Hepatol 2008. Jun;23(6):894–9. doi: 10.1111/j.1440-1746.2008.05420.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Maurovich-Horvat P, Massaro J, Fox CS, et al. Comparison of anthropometric, area- and volume-based assessment of abdominal subcutaneous and visceral adipose tissue volumes using multi-detector computed tomography. Int J Obes (Lond) 2007. Mar;31(3):500–6. doi: 10.1038/sj.ijo.0803454 [DOI] [PubMed] [Google Scholar]
- 25.Fox CS, Massaro JM, Hoffmann U, et al. Abdominal visceral and subcutaneous adipose tissue compartments: association with metabolic risk factors in the Framingham Heart Study. Circulation 2007. Jul 3;116(1):39–48. doi: 10.1161/CIRCULATIONAHA.106.675355 [DOI] [PubMed] [Google Scholar]
- 26.Bridge CP, Best TD, Wrobel MM, et al. A Fully Automated Deep Learning Pipeline for Multi-Vertebral Level Quantification and Characterization of Muscle and Adipose Tissue on Chest CT Scans. Radiol Artif Intell 2022. Jan 5;4(1):e210080. doi: 10.1148/ryai.210080 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Bridge CP, Rosenthal M, Wright B, et al. Fully-Automated Analysis of Body Composition from CT in Cancer Patients Using Convolutional Neural Networks. In: Stoyanov D, Taylor Z, Sarikaya D, et al, eds. OR 2.0 Context-Aware Operating Theaters, Computer Assisted Robotic Endoscopy, Clinical Image-Based Procedures, and Skin Image Analysis. CARE 2018, CLIP 2018, OR 2.0 2018, ISIC 2018. Lecture Notes in Computer Science, vol 11041. Cham, Switzerland: Springer, 2018; 204–213. doi: 10.1007/978-3-030-01201-4_22 [DOI] [Google Scholar]
- 28.Marquardt JP, Roeland EJ, Van Seventer EE, et al. Percentile-based averaging and skeletal muscle gauge improve body composition analysis: validation at multiple vertebral levels. J Cachexia Sarcopenia Muscle 2022. Feb;13(1):190–202. doi: 10.1002/jcsm.12848 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.McHugh ML. Interrater reliability: the kappa statistic. Biochem Med (Zagreb) 2012;22(3):276–82. Published October 15, 2012. Accessed September 13, 2022. [PMC free article] [PubMed] [Google Scholar]
- 30.Rigby RA, Stasinopoulos DM. Smooth centile curves for skew and kurtotic data modelled using the Box-Cox power exponential distribution. Stat Med 2004. Oct 15;23(19):3053–76. doi: 10.1002/sim.1861 [DOI] [PubMed] [Google Scholar]
- 31.World Health Organization. Statistical methods for constructing the growth curves. In: WHO child growth standards: length/height-for-age, weight-for-age, weight-for-length, weight-for-height and body mass index-for-age: methods and development World Health Organization, 2006. Pp.7–11 [Google Scholar]
- 32.Cole TJ, Green PJ. Smoothing reference centile curves: the LMS method and penalized likelihood. Stat Med 1992. Jul;11(10):1305–19. doi: 10.1002/sim.4780111005 [DOI] [PubMed] [Google Scholar]
- 33.Rigby RA, Stasinopoulos DM. Automatic smoothing parameter selection in GAMLSS with an application to centile estimation. Stat Methods Med Res 2014. Aug;23(4):318–32. doi: 10.1177/0962280212473302 [DOI] [PubMed] [Google Scholar]
- 34.van Buuren S, Fredriks M. Worm plot: a simple diagnostic device for modelling growth reference curves. Stat Med 2001. Apr 30;20(8):1259–77. doi: 10.1002/sim [DOI] [PubMed] [Google Scholar]
- 35.Royston P, Wright EM. Goodness-of-fit statistics for age-specific reference intervals. Stat Med 2000. Nov 15;19(21):2943–62. doi: [DOI] [PubMed] [Google Scholar]
- 36.Fuchs G, Chretien YR, Mario J, et al. Quantifying the effect of slice thickness, intravenous contrast and tube current on muscle segmentation: Implications for body composition analysis. Eur Radiol 2018. Jun;28(6):2455–2463. Published June 2018. Accessed June 17, 2023. doi: 10.1007/s00330-017-5191-3 [DOI] [PubMed] [Google Scholar]
- 37.Goodpaster BH, Park SW, Harris TB, et al. The loss of skeletal muscle strength, mass, and quality in older adults: the health, aging and body composition study. J Gerontol A Biol Sci Med Sci 2006. Oct;61(10):1059–64. doi: 10.1093/gerona/61.10.1059 [DOI] [PubMed] [Google Scholar]
- 38.Pickhardt PJ. Value-added Opportunistic CT Screening: State of the Art. Radiology 2022. May;303(2):241–254. doi: 10.1148/radiol.211561 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Adams SJ, Mikhael P, Wohlwend J, et al. Artificial Intelligence and Machine Learning in Lung Cancer Screening. Thoracic Surgery Clinics 2023. doi: 10.1016/j.thorsurg.2023.03.001. [DOI] [PubMed] [Google Scholar]
- 40.Adams SJ, Stone E, Baldwin DR, Vliegenthart R, Lee P, Fintelmann FJ. Lung cancer screening. Lancet 2023. Feb 4;401(10374):390–408. doi: 10.1016/S0140-6736(22)01694-4. [DOI] [PubMed] [Google Scholar]
- 41.US Preventive Services Task Force. Screening for Lung Cancer: US Preventive Services Task Force Recommendation Statement. JAMA 2021;325(10):962–970. doi: 10.1001/jama.2021.1117 [DOI] [PubMed] [Google Scholar]
- 42.Kanis JA, McCloskey EV, Johansson H, et al. A reference standard for the description of osteoporosis. Bone 2008. Mar;42(3):467–75. doi: 10.1016/j.bone.2007.11.001 [DOI] [PubMed] [Google Scholar]
- 43.Faron A, Sprinkart AM, Kuetting DLR, et al. Body composition analysis using CT and MRI: intra-individual intermodal comparison of muscle mass and myosteatosis. Sci Rep 2020. Jul 16;10(1):11765. Published July 2016. Accessed June 17, 2023. doi: 10.1038/s41598-020-68797-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.