Diagnostic and prognostic relevance of imaging-based body composition analysis in postmenopausal women: a review

Full Text

INTRODUCTION

The assessment of body composition (BC) involves quantifying and studying the distribution of body elements at various levels. The role of adipose tissue distribution in the incidence of metabolic diseases was first revealed in the late 1940s [1].

Among the different types of adipose tissue, the visceral adipose tissue (VAT), which is found in the thorax and abdomen and surrounding organs, has been identified to have a stronger association with cardiovascular disease (CVD) and metabolic syndrome (MS) [2, 3]. Studies have shown that VAT has high metabolic activity, secreting inflammatory markers into the portal circulation, which contribute to cardiovascular risk [2].

The presence of lipid deposits in other sites, such as skeletal muscle and the liver, has been linked to insulin resistance and metabolic risk phenotypes, independently of total adiposity. Recent investigations into marrow adipose tissue (MAT) showed an inverse relationship between MAT quantity and bone quantity and quality. In contrast, subcutaneous adipose tissue (SAT) has been hypothesized to play varying roles, from protective to contributing to insulin resistance, particularly at the abdominal level [4].

Anthropometric methods, such as waist circumference, hip circumference, waist-to-hip ratio, and body mass index (BMI), have long been used in clinical practice. Although easy to obtain and cost-effective, they have low accuracy and reproducibility [5]. These methods show poor correlation with VAT and SAT measurements from CT, which is the current gold standard for BC assessment [6]. Consequently, imaging has become crucial for accurate clinical evaluation of BC.

BC analysis is widely used in clinical practice for multiple purposes, including the assessment of cardiovascular and metabolic risk stratification, sarcopenia [7], osteoporosis [8], cancer cachexia in oncology [9], nutritional [10] and rehabilitation monitoring, and prediction of surgical outcomes and frailty [11]. Moreover, it is used across specialties including geriatrics, endocrinology, oncology, and critical care.

MECHANISMS OF AGING IN WOMEN

Menopause is defined by WHO as the permanent cessation of menstruation due to loss of ovarian follicular activity, marked by declining estrogen and progesterone levels, and increased FSH and LH levels. Menopause is confirmed after 12 months of amenorrhea. Lower estradiol levels induce systemic effects, increasing the risk of osteoporosis, cardiovascular issues, and changes in mood and mental health.

Furthermore, aging in women involves a decrease in the somatotropic axis (GH and IGF-1), called "somatopause", which leads to physiological changes, especially in the brain [12, 13]. These aging-related changes resemble, but differ from, those observed in growth hormone deficiency, including increased cholesterol and triglyceride levels, increased cardiovascular risk, muscle mass loss, decreased exercise tolerance, and decreased strength.

Vitamin D deficiency is common in aging. Vitamin D together with parathyroid hormone (PTH) regulates calcium absorption in a negative feedback loop. Low vitamin D levels, combined with decreased calcium intake and kidney function, can lead to secondary hyperparathyroidism, which causes sarcopenia and muscle weakness [14]. Additionally, PTH may affect skeletal muscle by increasing intracellular calcium and decreasing plasma phosphate, which alters muscle function. This mechanism may demonstrate the association between high PTH levels and increased risk of falls and frailty [15, 16].

Cardiovascular Risk

Menopausal transition is associated with increased risk of metabolic and cardiovascular diseases, mediated by a drop in circulating estrogen levels [17], and chronic inflammation. Postmenopausal women exhibit increased inflammatory markers such as C-reactive protein, interleukin-6 (IL-6), and TNF-α [18].

BC changes with menopause, including increased fat mass (FM) and visceral fat, which activate macrophages that secrete pro-inflammatory cytokines including TNF-α and IL-6 [19] — stimulating the synthesis of pro-inflammatory factors, such as C-reactive protein and ferritin. Moreover, non-bone lean mass (LM) decreases, which is more directly related to the effects of aging. In premenopausal women, visceral fat represents 5%–8% of the total FM. This percentage increases up to 15%–20% in postmenopausal women [20], with the subsequent increase in production of inflammatory markers involved in the development of cardiovascular risk [19] and are associated with mortality [21, 22].

Several studies have shown that postmenopausal women are at increased risk of developing abdominal obesity and MS. However only a few studies have investigated the association between years since last menstrual period, BC, and cardiometabolic risk factors in healthy women with normal weight [23–26]. Wang et al. [27] found that lean mass (LM) decreased with years since menopause, whereas FM and fat distribution were more closely related to age than time since the last menstrual period.

In women, identifying the difference between consequences of natural aging and those derived from menopause is challenging [28].

Although studies have explored the relationship between physical activity energy expenditure and BC changes in postmenopausal women, many have not accounted for years since menopause. Longitudinal studies have shown that physical activity decreases with age, but those who maintain or increase activity experience less changes in BC [29, 30].

Sarcopenia

According to the European Working Group on Sarcopenia in Older People (EWGSOP), sarcopenia is the progressive loss of skeletal muscle mass and strength. "Sarcopenic obesity" refers to the combined loss of muscle mass and increase in fat, which results in relatively higher FM compared to LM. Sarcopenia is a significant aspect of frailty in older adults [31–34]. Decreased muscle functionality is a diagnostic criterion [35]. Sarcopenia is associated with high morbidity, mortality, hospitalization rates, and loss of independence in daily activities [32, 33]. These changes are caused by fat infiltration (myosteatosis) in the skeletal muscle, which is a result of estrogen deficiency [31].

There are two main mechanisms behind myosteatosis: accumulation of intramyocellular lipids and disproportionate differentiation of mesenchymal stem cells into the adipogenic lineage, depositing fat between muscle fibers. Fat infiltration negatively affects muscle health and function, reducing insulin sensitivity [31].

Mood and Mental Health

Estradiol influences neurotransmitter systems, impacting mood and mental performance. Estrogens induce a neuroprotective effect through an antioxidant pathway; hence, decreased estrogen levels are linked to affective and cognitive disorders and increased risk of Alzheimer’s disease [36].

ASSOCIATION OF FRAILTY WITH BODY COMPOSITION

The diagnosis of frailty requires the existence of three of the following criteria: muscle strength loss, slowness, fatigue, low physical activity, and loss of body weight [37]. Physical frailty is considered a geriatric syndrome and has become a public health issue worldwide; it has been defined as a biological syndrome wherein a decrease in stress resistance and a decrease in different physiological functions lead to the inability to perform normal life activities and vulnerability resulting in increasing risk of falls, hospitalization, and adverse outcome as disability and death [37].

Women are more prone to develop fragility than men, showing from the earliest stages of loss of muscle mass and fat-free mass (FFM) and increase in total FM and skeletal muscle weakness [37].

Adipose tissue has metabolic functions, releasing cytokines such as IL-6 and TNF, which promote inflammation and insulin resistance, accelerating muscle mass and strength decline. Decreased skeletal muscle mass impacts several physiological functions, contributing to nutritional, functional, endocrine, and cognitive decline and increasing the risk of comorbidities in older adults [38].

However, the relationship between BC and frailty syndrome remains complex and unclear.

CURRENT STATUS OF IMAGING OF BODY COMPOSITION APPLIED TO AGING WOMEN

As women age, their BC changes, including height loss, decreased bone mineral density (BMD); leading to osteopenia and osteoporosis, increased FM, and decreased LM (sarcopenia and sarcopenic obesity). FM redistribution occurs particularly with an increase in central and VAT [39]. The present study focused on the analysis of BC, particularly FM and LM and their image quantification through the different available techniques. Assessment of intracellular fat content (e.g., in the liver and skeletal muscle) is also possible, with different imaging parameters linked to metabolic and cardiovascular risk. For extension purposes, a deeper insight on this specific topic is beyond the aim of this review.

Wang et al. [40] indicated a model of BC assessment based on complexity levels: atomic, molecular, cellular, organ-tissue, and whole body, each with different direct measurement methods, such as radioisotopic, biochemical, and anthropometric techniques.

Imaging techniques, such as DXA, ultrasound, CT, and MRI, are able to depict body components at different levels. DXA can distinguish between fat, water, proteins, and minerals at molecular level, and MRI is able to identify intracellular components. Nevertheless, the assessment of BC is mainly performed at tissue level with in vivo imaging techniques (Table 1) [12].

 

Table 1. Level of assessment of BC with different imaging techniques

Levels
Atomic	Molecular	Cellular	Tissue-organ	Whole-body
O	Minerals	ECS	Skeleton	—
	Glycogen		Visceral organs and residual
		ECF
	Proteins
	Water		Skeletal muscle
C
H		Cells adipocytes
	Lipid		Adipose tissue
N, Ca, P, R, Na, Cl
Medical imaging techniques
—	DXA fat + water + proteins + glycogen + minerals: soft tissue water + proteins + glycogen + minerals: lean soft tissue fat tissue bone minerals	US intracellular fat (muscle echogenicity) CT intracellular fat (muscle density) MRI intracellular fat (IMAT)	DXA VAT SAT US muscle thickness VAT-SAT cross-sectional area pennate angle fascicle length CT VAT-SAT cross-sectional area VAT-SAT volume SMI MRI VAT-SAT cross-sectional area VAT-SAT volume	DXA Total body LM FM/LM ALMI LMI SMI

 

Dual-energy X-Ray absorptiometry DXA is a reference standard technique for BC assessment, with advantages regarding accuracy, reproducibility, low cost, availability, and minimal radiation dose. DXA is the gold standard for defining BMD, and it is widely used to assess body fat and lean tissue mass at regional and whole-body levels [35].

DXA measurements are based on a three-compartment model, dividing body components into FM and FFM, with the latter further divided into LM and bone mineral content (BMC) (Fig. 1) [41].

 

Fig. 1. DXA analysis of body composition. Measurements are based on a three-compartment model that can be simplified into fat mass (FM: yellow), non-bone lean mass (LM: red), and bone mineral content (BMC: white). Body masses and bone mineral density (BMD) can be assessed on a regional or whole-body basis (Original image, created by the authors).

 

The technology relies on two different energy X-ray beams (low and high photon energy) transmitted through the body, whose attenuation depends on different tissue densities. The high and low energy peak attenuation quotient is the "R-value" and is thus tissue-specific.

Bone mass and FM R-values are considered constant, whereas soft tissue R-value may vary based on hydration status and fat infiltration [41].

R-value is assessed at anatomical regions containing both soft tissue and bone, and soft tissue only, adjacent to bone to quantify FM and fat-free soft tissues. When the R-value in these regions is obtained, FM, LM, and BMD can be differentiated in the reconstruction process. In particular, given a stated R-value, pixels wherein the R is below the threshold are classified as soft tissue, and R is linear to the fat tissue content (the higher the percentage of fat tissue, the lower the ratio) [41].

For pixels above the threshold, containing bone and soft tissue, an interpolation between soft tissue properties and surrounding soft tissue pixels is required [42–44].

A limitation of DXA in defining BC, especially LM, is related to patient hydration status. A constant hydration of FFM around 73% is assumed, with a variability ranging from 67% to 85%; thus, fluctuation due to food intake and exercise may affect DXA accuracy [45].

As people age, FM increases, whereas LM decreases (sarcopenia and sarcopenic obesity), and fat redistributes, particularly to central and visceral regions, which are associated with higher metabolic and cardiovascular risks [41].

DXA can quantify whole-body and regional FM and LM, including the trunk, arms, legs, and android and gynoid regions. DXA has been shown to be a better predictor of metabolic risk than anthropometric measures such as waist and hip circumference (Table 2) [46, 47].

 

Table 2. Summary of imaging markers and main applications

PARAMETERS	ADVANTAGE	DISADVANTAGE	USE
Dual-energy X-ray absorptiometry
Total body LM FM/LM ALMI LMI SMI ASMI VAT SAT	low cost low radiation exposure short acquisition time large availability	low accuracy for LM assessment for the constant hydration status assumption	Clinical settings for health surveillance and screening: fat mass assessment in patients with HIV using antiretroviral agents at risk of lipoatrophy fat and lean mass assessment in patients undergoing bariatric surgery fat and lean mass assessment in patients with muscle weakness and poor physical functioning
Ultrasound
muscle thickness VAT-SAT cross-sectional area pennate angle fascicle length intracellular fat (muscle echogenicity	low cost nonionizing radiation portable technique short acquisition time real time visualization	operator dependent no standardized protocol poor reproducibility	Clinical settings: fat assessment in patients undergoing bariatric surgery fat assessment in pregnant women fat and lean mass assessment in spinal cord injured patients
Computed tomography
VAT-SAT cross-sectional area VAT-SAT volume SMI intracellular fat (muscle density)	excellent spatial resolution high accuracy	ionizing radiation high cost no standardized protocol	Research settings Clinical settings: acquired exams for other clinical and therapeutic purpose
Magnetic resonance imaging
muscle volume/cross-sectional area VAT-SAT cross-sectional area VAT-SAT volume intracellular fat (IMAT)	high spatial and contrast resolution nonionizing radiation	high cost long acquisition time patient with noncompatible medical devices (pacemaker and metallic foreign bodies) no standardized protocol	Research settings Clinical settings: acquired exams for other clinical and therapeutic purpose

 

The android region in DXA analysis was designed to represent VAT and thus to predict metabolic risk. VAT accuracy to identify at least one laboratory risk factor associated with MS, among blood fasting glucose, total and HDL cholesterol, triglycerides, AST, and ALT, using ROC curves, was found high for men (AUC: 0.82; CI, 0.748–0.884) and women (AUC: 0.811; CI, 0.746–0.865) [48].

Furthermore, VAT values > 1395 g and > 1479 cm³ for men and >1281 g and >1357 cm³ for women, in Caucasian population, were pointed out as cutoff to identify at least one laboratory risk factor associated with metabolic syndrome. Most of the risk has been related to VAT compartment, whereas SAT plays a controversial role. The endocrine function of the two compartments is distinct, with different impacts on glucose metabolism [48].

Technically, because DXA provides a two-dimensional projection, a direct volumetric compartmental measurement of VAT and SAT is not possible; however, indirect estimation can be obtained from anatomical models [42]. The detection and quantification of ectopic fat is not possible with DXA.

A recently introduced DXA tool allows for the separate quantification of VAT and SAT in the android region (a segment of the abdomen comprised between an upper line drawn at a level representing 20% of the distance between the iliac crests and the chin and a lower line drawn joining the superior limits of the iliac crests). SAT can be estimated and then subtracted from the total FM in the android region to obtain VAT (as mass and volume). VAT measured using DXA has been validated against CT in a wide range of BMI values (18–40 kg/m²) and ages (18–90 years old), in both sexes [49].

The International Society for Clinical Densitometry (ISCD) stated that Z-scores for BC should be derived using the 2009 NHANES DXA data and representative for the US population [50, 51]. These have been integrated into Hologic and GE systems software to generate Z-scores for various adiposity and LM measurements [52, 53].

Recently, Ofenhemier et al. provided data regarding age- and sex-related reference values for BC parameters in European adults aged 18–81 years. These included VAT, FM index (FMI), android/gynoid and trunk/limbs FM ratios, LM index (LMI), and appendicular LMI (ALMI) [46].

Various studies have compared DXA and MRI, which is the gold standard in BC assessment. Measurements in both techniques can be compared after a linear transformation as they measure different entities (fat tissue and lean tissue mass estimation from areal measurement and volume, respectively) and there are differences in the assessed anatomical region [42, 54].

These studies showed an excellent agreement, good accuracy, and very good repeatability between the two methods. In particular, Borga et al. [42] reported excellent agreement for DXA and MRI in whole-body fat and lean tissue mass assessment, with a correlation coefficient of r = 0.99 for body fat and r = 0.97 for lean tissue, respectively, but lower agreement for VAT, particularly obese patients. Furthermore, the lowest correlation was found for arms FM assessment, due to signal loss on MRI.

Agreement between DXA and whole-body CT assessment of FM has also been found to be high, with a correlation coefficient of 0.99 [54].

LM, which is a crucial indicator of metabolic risk, decreases with age and is a key factor in diagnosing sarcopenia. Commonly, appendicular LM (ALM; the sum of LM from arms and legs) determined by DXA is considered proxies for skeletal muscle mass and a valid representation because it shows high correlation with whole-body imaging data in cross-sectional and longitudinal studies [55]. A correlation coefficient of 0.77–0.97 was found for whole-body and regional scans between ALM determined by DXA and muscle volumes derived from CT and MRI [56–63].

DXA’s high accuracy, availability, and low radiation (0.001 mSv) dose make it the most commonly used method for assessing skeletal muscle mass, although it assumes constant hydration of LM, which may decrease accuracy in some cases, especially in patients with edema [64].

In evaluating LM, and above all for sarcopenia diagnosis, several indices have been proposed during the years by different expertise groups, with consequent great heterogeneity [35, 65, 66].

DXA-derived indices can be divided by parameter of origin: skeletal muscle mass or LM lean muscle mass and parameter adjusted per height and weight for a greater comparability [65, 66].

In 1998, Baumgartner et al. [65, 66] suggested an operational definition of sarcopenia, intended as appendicular SM (ASM; kg) per height squared (m²) being 2D below the mean value in a young reference group.

Later in 2010, ASM and in particular appendicular SM index (ASMI: ASM/height²) have been recommended by the EWGSOP as parameter of sarcopenia diagnosis, adding low muscle function (strength or performance) to the presence of low muscle mass as new definition in the context of aging [35].

ALM (legs LM + arms LM), LMI (total LM/height²), and ALMI (ALM/ height²) [67] are used as SM indices by the Foundation for the National Institutes of Health Sarcopenia Project to determine low muscle mass [68]. Clearly, ALMI is a parameter of clinical significance, considering that the maintenance of appendicular skeletal muscle mass is critical for maintaining mobility and functional independence with aging, with subsequent impact on morbidity [69].

According to ISCD guidelines [51], LM status may be defined using ALMI with Z-scores obtained from age, ethnicity, and sex-matched population. Over the last decade, numerous consensus groups [55, 70–73] have proposed different ALMI sex- and age-related cutoffs from the general population in different countries, emphasizing the need to have different cutoffs for different ethnic groups; however, the threshold for the definition of low LM remains to be established and validated [51]. However, these data may be used as reference to monitor the decrease of muscle mass [74], given DXA indices used for LM are considered proxies for skeletal muscle mass.

The concept of sarcopenic obesity has made the definition of normal skeletal muscle mass even more complicated, as the presence of fat infiltration affects muscular quality and function [75].

Patients with low LM and high FM have higher morbidity and mortality than patients with only high FM [75]. Furthermore, increased energy expenditure and muscle force are required for muscular work in patients with obesity, because of the limitation of skeletal muscle strength induced by FM.

Thus, other parameters that combine muscle mass with FM and obesity have been proposed to facilitate understanding of the contribution of FM and LM to health risk such as ALM adjusted for BMI and % of body weight [76, 77].

In a previous study, the amount of LM in healthy women was constant from the third to eighth decade of life, whereas in the analysis by Kyle et al, lean mass gradually decreased in women aged > 34 years [39, 78]. Wang et al showed that FFM decreases with years since menopause in postmenopausal women, whereas FM and body fat distribution were not related to years since menopause but to age [27].

Currently, according to the official positions of ISCD [51], the application of DXA with regional analysis is limited to patients with muscle weakness and poor physical functioning to assess FM and LM, in patients with obesity undergoing bariatric surgery to assess FM and LM changes, and in patients with HIV to assess fat distribution.

ULTRASOUND

Ultrasound is a nonexpensive, portable technique that involves no ionizing radiation.

There are parameters and indexes of adipose tissue thickness that may be measured using conventional ultrasound imaging and have been proven to correlate with clinical and laboratory parameters (Fig. 2). However, standardization is required: intra-abdominal fat thickness, epicardial fat thickness, and peri- and pararenal fat thickness show good accuracy and reliability, with good correlation with areas and volumes obtained with CT and MRI [79]. In various studies, abdominal wall fat index, mesenteric fat thickness, and pre-peritoneal fat thickness demonstrate variable accuracy and reliability when compared with CT [79]. Other indexes, such as subcutaneous fat thickness, showed good correlation with areas observed using MRI and CT, and minimal and maximal subcutaneous fat thickness showed good correlation with areas observed using CT [79].

 

Fig. 2. Linear measurements of adipose tissue on ultrasound and CT. a, Minimal abdominal subcutaneous fat thickness (MinASFT) is the distance between the anterior surface of the linea alba and fat–skin barrier (1), obtained at a plane through the subxiphoid region. Pre-peritoneal fat thickness (2) is measured from the anterior surface of the peritoneum covering the liver to the posterior surface of the linea alba, at the plane through the subxiphoid region. b, Maximum abdominal subcutaneous fat thickness (MaxASFT) is the distance between the anterior surface of the linea alba and fat–skin barrier (3), measured at the level of the supraumbilical region. Intra-abdominal fat thickness is often measured as the distance from the posterior wall of the abdominal muscle to the anterior wall of the aorta in the supraumbilical region (4). (Original image, created by the authors).

 

Ultrasound can provide insights into intracellular fat, such as liver steatosis, which is associated with MS and CVD, and intramuscular fat, although its reliability varies (Table 2) [80].

Moreover, ultrasound can provide quantitative and qualitative analysis in muscle evaluation, through the parameters of muscle thickness, cross-sectional area (CSA), pennation angle, fascicle length, and echo intensity; the latter allows for the assessment of the presence of myosteatosis, fibrosis, and inflammation [79, 80].

Although CT and MRI are the gold standard for LM quantification, muscle mass estimates measured using ultrasound show good-to-high correlation with these methods [81]. However, none of the definitions of sarcopenia has included ultrasound in the diagnostic algorithm.

Muscle mass is estimated from muscle thickness measurements, using a specific equation [82].

In a population of healthy Japanese adults, Sanada et al. [82] measured muscle thickness in nine different sites and formulated regression-based prediction equations to estimate total and regional skeletal muscle mass, derived from the multiplication of muscle thickness by body height. When compared to MRI, a significant and strong site-matched correlation was found.

The main structural parameters related to muscle functional qualities measured by ultrasound in skeletal muscles, namely, fascicle length, pennation angle, and CSA, which is a parameter obtained from muscle volume divided by fascicle length, were found to be better related to muscle function than CT- and MRI-measured CSA [83].

The pennation angle is formed at fibers attachment site into deep and superficial aponeurosis and can be evaluated in static and dynamic conditions, providing information about contractions and mechanical properties [84].

Alongside the advantage of the dynamic evaluation of the pennation angle, there are limitations related to the use of 2D images with limited FOV and inaccurate measurement which may derive from a malalignment of the transducer, [84, 85] resulting in poor image quality, especially for the assessment of deeply located muscles.

MRI can overcome these limitations, but lacks resolution to demonstrate individual muscle fascicles. Recently, a method based on diffusion tensor imaging was considered as reference method to quantify muscle architectural parameters in vivo [86].

The general limitations of ultrasound for BC analysis include operator dependency and lack of standardized protocols, which reduce reproducibility. Additionally, all parameters measured by ultrasound are influenced by aging to varying degrees, and further validation is warranted (Table 2).

COMPUTED TOMOGRAPHY

Computed Tomography (CT) technology has evolved to offer quantitative modalities to study bone and muscle. Currently, CT and MRI are the gold standard for investigating BC [35].

CT allows for quantification of adipose tissue, its distribution, and skeletal muscle composition in different body segments with high accuracy [87].

The modern multislice CT scans with their helical acquisition enable obtaining 2D and 3D high-resolution images of the whole body or parts of the body. The differentiation between FM and FFM is based on different tissue X-ray attenuation, which results in specific HU values. Negative Hounsfield unit (HU) values conventionally identify adipose tissue; different ranges were proposed from a minimum value of −200 up to −30 [88], with a slight difference found between VAT and SAT (−150 to −50 and −190 to −30, respectively).

The HU range for muscle varies from the highest HU value for adipose tissue (−30) up to 150 [89, 90].

In the clinical setting, to reduce radiation exposure, BC analysis by CT is limited to 2D analysis of a selection of axial images or a single axial slice, using the measured areas as a proxy to estimate the total volume of FM and LM.

Single-slice CT-based BC measurement showed good agreement with single-slice MRI measurement [42, 91–93].

The most widely used level is the third lumbar vertebra (L3), followed by C3 [94], T12 [95], and thighs at midfemur level. The most frequently used parameters are CT-derived CSA and volumetric estimates, which showed good correlation with cadaver studies [91].

Quantitative 2D and 3D CT BC analysis requires segmentation of tissue compartments, utilizing computer-assisted diagnostic tools. These range from manual to fully automated segmentation software. Semiautomatic methods are most common, providing a balance between accuracy and efficiency [90].

The most commonly used software is the semiautomatic, wherein a threshold-based segmentation is completed by human analysis correction, overcoming the time-consuming drawback of the task [96].

Furthermore, CT-quantitative tools for BC assessment can be classified into two groups based on access/ availability: commercial and open-access. Available open-access software (e.g., 3D Slicer, ITK Snap, MITK, ePAD Imaging Platform, ImageJ, and RIL-Contour) has the advantage of free accessibility, whereas the commercial ones (e.g., SliceOmatic, OsiriX, and AMBRA) are customizable for imaging systems [90].

CT-derived adipose tissue quantification parameters, such as VAT and SAT area and volume, correlate with clinical parameters, although these measurements have no standardized cutoff values (Fig. 2).

CT also quantifies muscle mass and density. Muscle mass is typically measured using CSA and normalized for patient height to derive the skeletal muscle index (SMI) [97]. Muscle density, which is used to quantify myosteatosis, is assessed using intramuscular adipose tissue (IMAT) area and muscle attenuation (MA) [98].

The association between these two parameters improves the evaluation of sarcopenia as a combination of muscle loss and impaired function, to which myosteatosis may contribute [97, 99–104].

SMI is most commonly assessed at muscle groups of the abdominal wall, at the L3 level, whereas IMAT and MA are more frequently assessed at mid-thigh level. In a recent review, Amini et al. indicated that the most common cutoff ranges for SMI at abdominal wall muscles are 52–55 cm²/m² for men and 39–41 cm²/m² for women. However, there is no consensus, and no standardized diagnostic cutoff values are currently used for other skeletal muscles [97].

Nonetheless, CT has limitations, such as high radiation exposure (approximately 8 mSv for a standard CT scan), which is significantly higher than natural radiation (2.5 mSv) [64].

New methods, such as quantitative CT (QCT) and peripheral QCT (pQCT), enable analysis of both bone and soft tissue parameters, respectively, at central sites and peripheral locations [105].

Aside from BMC, pQCT may also allow for the quantification of soft tissue areas and is currently being used to investigate the relationships between bone, muscle, and fat [105].

Starting from analysis limited to the tibia and forearm, technological developments and the use of larger machines enable the analysis of wider areas, including those up to the mid-thigh level. Furthermore, pQCT is able to quantify muscle density and cross-section area with good precision, becoming a frequently used method to measure skeletal muscle mass at 66% of the tibia [105].

The HR-pQCT (high-resolution peripheral quantitative CT), which is the pQCT modern evolution more recently introduced, is becoming the main modality in quantifying bone microarchitecture at the distal tibia; radio with higher resolution than pQCT and the second-generation models with larger gantry than the first-generation ones, which are able to scan only distal parts of the limbs, can acquire images at the 66% proximal tibia site, allowing assessment of muscle parameters with higher resolution (60.7 μm voxel size compared to 200 μm) [105].

Erlandson et al. first compared muscle CSA and muscle density measured with pQCT with myotendinous tissue CSA and density obtained using HR-pQCT in postmenopausal women, revealing moderate correlation (r = 0.44; p < 0.001) for CSA and muscle density (r = 0.69–0.70; p < 0.01) [106].

Later, Hildebrand et al. obtained positive correlation from the comparison of compared muscle CSA and density obtained with pQCT and second-generation HR-pQCT (R² = 0.66, R² = 0.95, respectively; p < 0.001) [107].

Furthermore, with its high resolution, HR-pQCT is able to discriminate fat tissue from lean skeletal tissue, allowing the definition of muscular quality parameters such as IMAT. New methods and protocols are being developed with the aim to obtain information on bone health and identify myosteatosis and sarcopenia in the early phase for a prompt and personalized therapeutic approach, with rapid and low-radiation-dose HR-pQCT scan (3μSv per scan) [108].

MAGNETIC RESONANCE IMAGING

Magnetic Resonance Imaging (MRI) is a cross-sectional imaging technique that provides 2D and 3D images with high tissue contrast resolution and high spatial resolution, representing the gold standard for in vivo assessment of BC [64].

In particular, MRI can determine the amount and distribution of adipose tissue in different subgroups VAT, SAT, and ectopic fat divided in the muscle, liver, pancreas, heart, and bone marrow [109, 110]. Additionally, MRI can assess muscle volume and its qualitative composition, detecting early fat replacement with higher sensitivity than CT [64, 111]. MRI BC analysis involves several steps: image acquisition, reconstruction, segmentation, and quantification [112]. Several sequences to identify and measure body fat and many different methods of imaging acquisition have been developed; however, a standard study protocol remains to be established. T1-weighted MRI sequences, which show high signal intensity in adipose tissue, make it useful for quantifying SAT, VAT, bone marrow fat, and IMAT (Fig. 3). T2-weighted imaging, with fat tissue hyperintensity, is mainly used for fat quantification in lower-extremity muscles, but is unsuitable for SAT and VAT assessment [113].

 

Fig. 3. MRI areal measurement of the different fat compartments. (Original image, created by the authors). a, T1 FSE image through the L2–L3 level. b, Segmentation of SAT (orange), VAT (yellow), and non-adipose tissue (blue). The gas within bowel loops is seen as black and the bone as white.

 

For ectopic fat localized in the skeletal muscle (further divide in intra- and extra- myocellular lipids, IMCL, and EMCL), liver, bone marrow, and heart, the single-voxel ¹H-based MR spectroscopy (H-MRS) is the gold standard noninvasive in vivo assessment method, using point resolved spectroscopy or stimulated echo acquisition mode sequences [114].

Fat quantification can be achieved with high spatial resolution using chemical shift encoding-based water-fat MRI. This is an advantage when compared to single-voxel MRS because the fat content can be heterogeneous in space, especially in the bone marrow [110, 115].

This method, also called the Dixon method, codifies during a single acquisition, chemical shift, and spatial resolution and assesses water and fat contribution to each voxel signal assuming a fixed offset in resonance frequency between water and fat. As water and fat molecules process at different rates, alternating between in-phase and out-of-phase, acquiring in-phase and out-of-phase images creates four combinations: water only, fat only, water and fat, and water minus fat [115].

This technology may detect focal and diffuse fat ectopic accumulation, especially in the liver, to separate inter- and intramuscular fat assessment, overcoming the limitations of T1-weighted imaging, and to quantify VAT and SAT and obtain separate water and fat images [113].

Different methods of imaging acquisition have been developed from whole-body acquisition to single-slice and multislice region-specific protocol.

Whole-body scanning is the most precise in defining body fat distribution and quantification; however, it is time-consuming and mainly reserved for research studies [116].

The most used is abdominal region scanning, using two protocols: multislice and single-slice acquisition. The former is the method of choice for longitudinal studies; the acquisition from the top of the liver to the top of the femoral head with continuous or separate slices provides full information of BC in less time [116].

Single-slice protocol acquisition is emerging as a viable option, especially for large-scale studies, because of its fast acquisition, relative simplicity in post-processing and segmentation procedure, and good correlation with fat and lean tissue measurement with multislice acquisition and whole-body scanning [116].

However, there is no consensus on a standard anatomic landmark, owing to sex, age, BMI and anthropometric characteristics variability. Research shows that a single-slice acquisition at the L3–L4 level provides the best correlation with whole-body SAT, whereas L1–L2 correlates best with VAT [117–121].

Several studies have recommended a good prediction of total skeletal muscle volume, VAT, and SAT through single-slice MRI acquisition at L3 level [122–124].

Kuk et al. [125] showed that in postmenopausal women, VAT assessment by single-slice acquisition at L1–L2 vertebral level has the strongest correlation with MS parameters.

The complexity of data analysis has led to the development of semi- or fully automated systems for segmenting body components, using signal intensity histograms and tools such as edge detection [126].

Automatic VAT assessment by T1w and water–fat imaging methods demonstrates comparable diagnostic accuracy [127], with VAT being associated with more severe metabolic, atherogenic, dyslipidemic, and obesity phenotype compared to the amount of SAT and serving the strongest determinant of insulin sensitivity [128].

However, West et al. showed that in postmenopausal women, Dixon sequences are able to characterize a wider range of fat and muscle compartments with high precision, such as VAT, SAT, and muscle infiltration in the thigh, lower leg, and abdominal muscles, with a coefficient of variation of 1.1–1.5 for VAT and SAT and 0.8–1.9 for the muscle groups [129].

MRI parameters have been correlated with metabolic control and diet and physical activity intervention in diabetes [130, 131].

Although MRI is the gold standard for muscle quality and mass research, its high cost, limited availability, and complexity restrict its clinical use. MRI quantifies muscle size (thickness, CSA, and mass) and quality (fiber length and pennation angle) (Table 2) [91].

Again, no consensus has been reached regarding the level of acquisition and specific cutoff to assess SM and to classify the loss in the sarcopenic range.

Single-slice acquisition at L3 level provides an estimation of total abdominal muscle area; however, most protocols concentrate on mid-femur-level acquisition, representative of thigh skeletal muscle and adipose tissue infiltration [132].

One of the advantages of MRI is the ability to detect changes occurring with aging and disease progression in the muscle structure, particularly the progressive accumulation of adipose tissue and fibrous connective tissue (both noncontractile tissues) and development of edema [133].

CSA values (cm²) and the volume of adipose tissue-free skeletal muscle, adipose tissue surrounding muscles, and adipose tissue contained within muscles (interstitial adipose tissue) obtained with MRI have been found to have an excellent correlation with values from cadaveric studies [91].

Chemical shift-based water–fat separation and 2-point Dixon-based and H-MRS can quantify the amount of intramyocellular lipid, which has been found to negatively correlate with insulin sensitivity [114].

In metabolic studies, IMAT is considered a distinct depot contributing to MS, independent of VAT [121].

Macaluso et al. [134] reported a significantly larger amount of intramuscular noncontractile tissue and a significantly lower amount of contractile muscle volume in older women (in the postmenopausal bracket, mean age: 69.5 ± 2.4 years) than in younger women (mean age: 22.8 ± 5.7 years), in hamstrings and quadriceps. However, a major issue with MRI is the lack of standardized protocols, hindering study comparison. MRI methods are primarily used in research to understand metabolic diseases such as MS, obesity, and type 2 DM [114, 131]. Although promising, these methods are still experimental for clinical practice.

The future of BC analysis lies in artificial intelligence (AI), which can automate complex tissue segmentation and quantification, facilitating large cohort studies and clinical use. AI, especially deep learning with neural networks, is being developed to improve the speed and efficiency of MRI and CT-based analyses, addressing the practical limitations of manual segmentation in large-scale studies [135].

CONCLUSION

BC changes with every aspect of aging in women, particularly being affected by the endocrinological changes occurring with menopause.

Multiple BC imaging markers obtained through different techniques have been proposed and investigated, with a range of associations proven with cardiometabolic risk and with impact on morbidity/mortality, functional impairment, and frailty. The current level of evidence in the literature on reliability and potential associations of those imaging markers helps to determine which ones are stronger and worth to be applied in clinical practice.

The available imaging tools (DXA, ultrasound, CT, and MRI) can be used both on dedicated and opportunistic settings, with promising potential future uses.

The different BC markers obtained by imaging techniques may play a critical role in improving management aspects on aging, postmenopausal women, to prevent or decrease unhealthy aging and the rate of women with aging-related health complications.

ADDITIONAL INFORMATION

Author contributions: M.P. Aparisi Gómez: conceptualization, data curation, writing — original draft; M.R. Petrera: data curation, writing, reviewing; A. Santoro, M.L. Petroni, C. Franceschi, G. Marchesini, G. Guglielmi, A. Bazzocchi: writing — review & editing; C. Gasperini: data curation, writing — review & editing. All the authors approved the version of the manuscript to be published and agreed to be accountable for all aspects of the work, ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Ethics approval: Not applicable.

Funding sources: No funding.

Disclosure of interests: The authors have no relationships, activities, or interests for the last three years related to for-profit or not-for-profit third parties whose interests may be affected by the content of the article.

Statement of originality: No previously published material (text, images, or data) was used in this study or article.

Data availability statement: The editorial policy regarding data sharing does not apply to this work.

Generative AI: No generative artificial intelligence technologies were used to prepare this article.

Provenance and peer review: This article was submitted unsolicited and reviewed following the standard procedure. The peer review process involved two members of the Editorial Board.

About the authors

Maria P. Aparisi Gómez

Te Toka Tumai Auckland (Auckland District Health Board); Waipapa Taumata Rau — University of Auckland

Email: pilar.aparisi@tewhatuora.govt.nz
ORCID iD: 0000-0002-6483-7139
New Zealand, Auckland; Auckland

Miriana R. Petrera

National Institute for Infectious Disease “Lazzaro Spallanzani”

Email: mirianapetrera@gmail.com
ORCID iD: 0000-0002-1275-6265
Italy, Rome

Aurelia Santoro

University of Bologna, Sant'Orsola-Malpighi Hospital

Email: aurelia.santoro@unibo.it
ORCID iD: 0000-0002-7187-1116

MD, PhD, Associate Professor

Italy, Bologna

Maria L. Petroni

University of Bologna, Sant'Orsola-Malpighi Hospital

Email: marialetizia.petroni@unibo.it
ORCID iD: 0000-0002-7040-6466

MD, Associate Professor

Italy, Bologna

Chiara Gasperini

IRCCS Istituto Ortopedico Rizzoli

Email: chiara.gasperini@unibo.it
ORCID iD: 0000-0002-5306-0985
Italy, Bologna

Claudio Franceschi

University of Bologna, Sant'Orsola-Malpighi Hospital

Email: claudio.franceschi@unibo.it
ORCID iD: 0000-0001-9841-6386

MD, Professor

Italy, Bologna

Giulio Marchesini

University of Bologna, Sant'Orsola-Malpighi Hospital

Email: giulio.marchesini@unibo.it
ORCID iD: 0000-0003-2407-9860

MD, Professor

Italy, Bologna

Giuseppe Guglielmi

University of Foggia; “IRCCS Casa Sollievo della Sofferenza” Hospital

Author for correspondence.
Email: giuseppe.guglielmi@unifg.it
ORCID iD: 0000-0002-4325-8330

MD, Professor

Italy, Foggia; San Giovanni Rotondo

Alberto Bazzocchi

IRCCS Istituto Ortopedico Rizzoli

Email: abazzocchi@gmail.com
ORCID iD: 0000-0002-2659-4535
Italy, Bologna

References

Supplementary files