Imaging modalities for measuring body composition in patients with cancer: opportunities and challenges

Abstract

Body composition assessment (ie, the measurement of muscle and adiposity) impacts several cancer-related outcomes including treatment-related toxicities, treatment responses, complications, and prognosis. Traditional modalities for body composition measurement include body mass index, body circumference, skinfold thickness, and bioelectrical impedance analysis; advanced imaging modalities include dual energy x-ray absorptiometry, computerized tomography, magnetic resonance imaging, and positron emission tomography. Each modality has its advantages and disadvantages, thus requiring an individualized approach in identifying the most appropriate measure for specific clinical or research situations. Advancements in imaging approaches have led to an abundance of available data, however, the lack of standardized thresholds for classification of abnormal muscle mass or adiposity has been a barrier to adopting these measurements widely in research and clinical care. In this review, we discuss the different modalities in detail and provide guidance on their unique opportunities and challenges.

Body composition, defined as the proportion, distribution, and quality of muscle and adipose tissues in the body, is an important biomarker and endpoint in the care of patients with cancer. Muscle and adipose tissue are active, endocrine organs that affect metabolism, immune surveillance, and inflammation, as well as physical function, nutritional status, and quality of life (1-4).

There have been numerous studies linking measures of body composition to metabolic diseases. As an example, increased visceral fat is associated with an increased risk of hypertension, impaired fasting glucose, diabetes mellitus, and metabolic syndrome (5-9). However, cancer is also a metabolic disease; the deregulation and reprogramming of cellular metabolism is an established hallmark of cancer (10). Therefore, body composition and its effects on systemic metabolism may impact cancer development, progression, and outcomes.

In fact, evidence supports the association between specific body composition metrics with cancer-related outcomes, including patient-reported quality of life, dose-limiting toxicities, postoperative complications, hospitalizations, disease recurrence, disease progression, cancer-related death, and overall mortality (11-14). Sarcopenia, or low muscle mass, is associated with increased toxicity and worse overall and cancer-specific survival (15-17). This is compounded further by increased frailty associated with cancer treatments including radiation and chemotherapy in the curative and palliative settings (18,19). Furthermore, increased visceral adiposity is also associated with poor outcomes (20-22). Patients with cancer with a synergy of these poor prognostic phenotypes (ie, sarcopenic obesity) may be at even higher risk for mortality and severe complications from cancer treatment (11,23,24). These conditions are distinct from cachexia, which is characterized by extreme weight loss and muscle wasting (25,26). Cachexia is common (as high as 80%) among individuals diagnosed with advanced stage cancers and associated with poor prognosis (25,27).

It is important to note that these parameters cannot be assessed by routine body weight alone; further clinical, imaging, and analytic techniques are needed and may have an important impact on cancer patient management. Herein, we review and compare the current modalities available to assess body composition in patients with cancer.

Traditional markers of body composition

The standard and commonly used measures for body composition rely on using anthropometric measurements including body weight, height, and body circumference; these measures are used to derive the body mass index (BMI, kg/m²), waist circumference (WC), and waist-to-hip ratio (WHR). Body composition has also been estimated using skinfold (SFT) thickness and bioelectrical impedance. Given their simplicity, BMI, body circumference, and SFT thickness can be used quickly and easily in routine clinical settings. Ease of these measurements has allowed for comparison across datasets and populations, increasing applicability in epidemiologic research studies. However, each measure has unique limitations complicating its ability to measure true body composition.

Body mass index

BMI is one of the most widely accepted markers for body composition and has been used to diagnose and assess clinical outcomes of various diseases. It is calculated by dividing the weight (kg) by the square of the height (m²). It is primarily used for categorizing obesity (underweight BMI <18.5, normal weight 18.5-24.9, overweight 25-29.9, and obesity ≥30) (28-33). The major limitation of using BMI is measuring weight, not fat or skeletal muscle; additionally, it does not take into consideration other factors that might impact the body weight composition, such as sex, age, muscle mass, fat percentage and distribution, and nutritional or athletic status. In addition, because BMI categorization has been adapted from the National Institutes of Health and the World Health Organization guidelines (32,34-36), it may not be uniformly informative for all populations. It might, for example, underestimate body fat in some ethnic groups who may have higher fat percentages, such as Asians and Indians, and older adults. On the other hand, it may underestimate body fat in other ethnic groups who might have higher muscle mass (such as Black individuals) (34,37-39), athletes (40), or patients who have water retention and edema (41,42).

Body circumference

WC and WHR are popular measurements for central or truncal fat. In general, as WC and WHR increase, the risk for chronic diseases including cardiovascular diseases and type 2 diabetes also increases (32). The WHR is calculated by dividing the WC by the hip circumference. The cutoff points for increased risk of chronic disease according to the World Health Organization is a WC of more than 40 inches (102 cm) for males and more than 35 inches (88 cm) for females (43). For WHR, cutoff thresholds of at least 0.90 for males and at least 0.85 for females predict a substantially increased risk of metabolic complications (32,44). WC and WHR measurements are subject to error, thus training is required to ensure precision; all body circumferences are usually taken by measuring tape, which involves a high probability of tape or measuring protocol errors (45). WC and WHR may be affected by body positioning and body composition (bony landmarks can be difficult to palpate in obesity or other conditions). For example, to take an accurate measure, the person must stand and do an exhalation while the WC measurement is taken (30,43). WC is more sensitive than WHR in estimating truncal obesity, but both WC and WHR have similar limitations to BMI because they do not consider age, sex, or ethnicity (46-48). Other limitations include an inability to differentiate between subcutaneous fat and visceral fat depots that may have opposing effects on diseases such as diabetes (49,50). In addition, the WC may not reflect increased abdominal fat, and a large hip circumference may hide abdominal obesity using WHR (43).

Skinfold thickness

SFT is a technique that measures a double fold of skin with subcutaneous fat in various sites of the body, measured in millimeters using calipers (51). The usual measurement sites include triceps, biceps, subscapular, suprailiac, midcalf, and anterior abdominal wall. SFT is a simple, reliable, cheap, and noninvasive method that can be used for all ages (51,52), and race-specific equations can be used to calculate body fat percentage (53,54). However, similar to circumference measurements, accurate SFT measurements require adequate training and high-level skills, and the accuracy decreases in severe obesity (55).

Bioelectrical impedance analysis (BIA)

BIA estimates body composition by running a small electrical current through the body and measuring the resistance provided by fat and muscle (56-58). Fat-containing tissues will have less water that will slow or impede the electrical current, and muscle that contains more water will allow the current to progress quickly through the body (56,59). BIA is a simple, relatively inexpensive, quick, and noninvasive technique that also accounts for sex differences in body composition (56,59) and might be valid for different racial and ethnic groups (60). There are some reports that BIA can also be used in the assessment of sarcopenia. A systematic review of this topic identified that in some cases, sarcopenia identified by BIA was associated with poor clinical outcomes (61). However, the BIA results can be affected by hydration, food and alcohol consumption, menstrual cycle, electrode location, and exercise up to 24 hours prior to the measurement (59,62). Although newer multifrequency BIA technology is more promising and may have more significant correlations with computed tomography (CT) scan–based body composition measurements (63), the accuracy and clinical value of the BIA results are debatable and need further validation with dual x-ray absorptiometry (DEXA), CT, or magnetic resonance imaging (MRI) (25,62,64,65).

Quantitative imaging markers of body composition

The evolution of body composition assessment has progressed beyond the aforementioned methods to include imaging-based quantitative approaches including DEXA, CT, MRI, and positron emission tomography (PET). These imaging modalities, especially CT, MRI, and PET, are routinely used clinically to measure cancer burden in patients; the same scan obtained for clinical purposes can therefore be used to assess the body composition of these patients (66) (Table 1). Although the precision of imaging-based approaches is well established for use in clinical medicine, the question remains as to which method is ideal to assess body composition given the challenges and opportunities afforded by each modality.

Table 1.

Comparison of various modalities that measure body composition^a

Factor associated with imaging	DEXA	CT	MRI	PET
Scan cost	+	++	+++	+++
Radiation exposure	+	++	−	+++
Full-body assessment	+++	++	+	++
Patient access and throughput	+	+++	+	+++
Organ and compartment-specific analyses	−	+++	+++	+
Scan time	+	++	+++	+++
Tissue metabolism	−	−	++	+++
Organ anatomy	−	+++	+++	+

^aA qualitative grading scale comparing different modalities is given. “-” : absent; “+” : present, where “+++” > “++” > “+”. CT = computed tomography; DEXA = dual x-ray absorptiometry; MRI = magnetic resonance imaging; PET = positron emission tomography.

Dual energy x-ray absorptiometry

DEXA is a radiological method currently recommended to measure body composition in clinical care. Since the introduction of DEXA into clinical practice in 1987, DEXA has become the internationally accepted standard of care for measuring bone mineral density (BMD) and diagnosing conditions such as osteoporosis, which can lead to increased risk of falls and fracture (67-70). As technology evolved, DEXA was increasingly recognized as a method to measure additional body elements such as fat and lean muscle mass; these are critical measures among patients with cancer at increased risk for obesity, frailty, sarcopenia, and cachexia (25) (Figure 1).

Figure 1.

Analysis of body composition with DEXA. DEXA scan of 2 patients distinguishes between fat, lean, and bone mass. A) Female aged 66 years with BMI = 21.1 (WHO classification = normal) is calculated to have 34.4% body fat (ninth percentile for age-matched individuals). B) Female aged 52 years with BMI 33.6 (WHO classification = obesity I) is calculated to have 54.6% body fat (99th percentile for age-matched individuals). BMI = body mass index; DEXA = dual x-ray absorptiometry; WHO = World Health Organization.

DEXA technology is based on the principle that x-ray beams with 2 energy levels are attenuated differentially as they pass through body tissues and structures with different density (71,72). Essentially, the higher x-ray beam is attenuated less when passing through tissue relative to the lower x-ray beam, which is attenuated more prior to reaching the sensor (68). Based on this attenuation, the sensors produce the R value, which is the ratio of the attenuation coefficients for each energy level (68). The R value is constant for bone and adipose tissue across individuals but varies for soft tissue. As with any radiological method using x-rays, the radiation dose given to patients should always be considered. Whereas the effective dose of an anterior-posterior spine DEXA is approximately 10 µSv, newer methods for total body DEXA only expose patients to 4-5 µSv, which is lower than the daily natural background dose of 6.7 µSv radiation (68,71). The newer methods incorporate multiple sensors, including fan densitometers, as opposed to previously used pencil densitometers with a single detector. This advance in the DEXA technology allowed for improved images and shortened scan times, as well as lower radiation dose (67).

Based on the R values described above, DEXA produces measurements using a 3-compartment model to provide estimates of 1) bone, 2) adipose tissue, and 3) lean soft tissue mass (67,71). In assessing BMD, DEXA provides measurements of bone quantity, quality, and geometry (73). However, DEXA is a 2D measurement that cannot assess volume without additional statistical models applied (74), unlike CT or MRI as described below (69). DEXA produces a T score, used to compare the measured BMD to the young adult mean (specific for age and sex) vs the Z score, which measures the difference between the measured BMD and age-matched mean BMD (specific for sex and ethnicity). For muscle mass, DEXA produces measurements including appendicular skeletal mass and total body skeletal muscle mass but also measures from specific points such as the thigh, which is used as a measure of sarcopenia (26). Measures of adipose tissue include subcutaneous adipose tissue (SAT) and visceral adipose tissue (VAT) (68,75). These DEXA measurements provide a comprehensive and detailed view of body composition; however, DEXA is routinely used in clinical care only as a measure of bone density and seldomly used outside of research for muscle or adipose composition measurement.

Although DEXA is widely used in clinical settings for BMD and not body composition, there are limitations to the technology. DEXA measures are particularly sensitive to 3 areas including subject preparation, subject positioning, and scan postprocessing (68). One particular challenge is that DEXA can be influenced by the hydration status and digestive tract content of patients, which can lead to overestimation of lean mass and underestimation of fat mass (71). Reliability of DEXA may also vary by trunk thickness as well as lower accuracy when examining changes in body composition over time, relative to other methods (76). Thus, intertechnician error or inaccuracies in body positioning or image analysis can impact the quality and reliability of the DEXA measures. In fact, higher fat composition decreases DEXA reproducibility in bone density measurements (77). However, the limitations of DEXA are balanced by the fact that DEXA can be used widely, because the only clinical contraindication for DEXA use is pregnancy (72,76).

Computed tomography

CT imaging involves x-ray beam transmission from an x-ray tube source to a detector on the other side of the patient (78). CT imaging technology has been used clinically since the 1970s and, over time, has evolved to have faster speed, improved image quality, and lower radiation doses (78-80). In the most recent versions of CT, multiple tubes and detectors (ie, multidetector CT) are built into a ring configuration and are rotated around the patient as the patient is moved through on a sliding bed. This results in a full 360-degree scan that could, in theory, extend from the head to the toes, with the actual scanned body part dependent on the clinical question being asked (78). Although the specific image that is used to characterize body composition is dependent on the underlying indication for CT, abdominal CT is the most commonly used for body composition assessment (81) (Figure 2).

Figure 2.

Analysis of body composition with CT. A) An automated 3D segmentation algorithm (Data Analysis Facilitation Suite v. 3.7; Voronoi Health Analytics; Vancouver, BC) is applied to a CT dataset to compute volumes of multiple organs in the body (different colors) including muscle, bone, and the different fat compartments. B) Volumetric segmentation can also be applied on a slice-specific basis, such as the L3 vertebral level. Fat stores (subcutaneous fat, visceral fat, and intermuscular fat), skeletal muscle, bone, and multiple viscera (eg, lungs, liver, spleen, kidneys, and aorta) are segmented and color coded for volumetric quantification. CT = computed tomography.

The attenuation of x-ray beams as they pass through various tissues is calculated, expressed in Hounsfield units (HU), and ultimately used to reconstruct images (78,79). Adipose tissue does not attenuate x-rays well and therefore has lower HU compared with solid organs including muscle. In general, studies have used thresholds from −29 to +150 HU to measure skeletal muscle, −190 to −30 HU to measure subcutaneous and intramuscular adipose tissue, and −150 to −50 HU to measure VAT (−150 to -50 HU) (79,82,83). As a result, CT imaging is considered a precise, gold standard approach to estimating muscle mass (84). In fact, studies have found close correlations between CT and MRI measurements of adipose tissue, bone, and muscle, and direct measurements of these tissues in cadavers (81,85-87).

As a relatively fast and readily available imaging modality capable of providing detailed radiographic information, CT imaging is obtained frequently in the clinical setting. This is especially relevant in the context of the oncology population given the need for high-resolution and full-body imaging to evaluate the cancer at initial staging, for disease monitoring, and during surveillance. CT images have been applied to evaluate the association between body composition and treatment outcomes in patients across various oncologic diagnoses (eg, solid tumors, hematologic malignancies) (88-90). Additionally, serial imaging enables longitudinal assessment of changes in body composition following treatment that cannot be captured by conventional anthropomorphic measurements (eg, BMI) (82,91).

One key example highlighting the value of CT-based body composition in predicting patient outcomes is the measurement of myosteatosis, or fat deposition in muscle. In CT, myosteatosis can be estimated by muscle attenuation or density (measured in HU) where muscle with higher fat fraction has lower density. There is growing recognition that myosteatosis is an important predictor of outcomes in patients with cancer. A meta-analysis identified that patients with lymphoma, gynecologic, renal, pancreatic, hepatocellular, gastroesophageal, and colorectal cancers who had higher myosteatosis had worse overall survival (92). The conversion of CT muscle attenuation values to an actual muscle fat fraction has been attempted but has required the use of MRI (described below) (93).

There are, however, limitations to utilizing CT images to evaluate body composition. CT may not be routinely collected for disease monitoring in all cancer patient scenarios, such as those with early stage breast cancer or hematologic malignancies (94). In addition, the associated costs and radiation exposure could preclude obtaining images solely for body composition measurements outside the realm of clinical care. The nature of the CT scanner limits its portability, which is an important consideration in designing larger, scalable studies (81). However, this is balanced by the ubiquitous presence of CT scanners in health-care facilities as well as the development of portable CT scanners (95).

Magnetic resonance imaging

There has been increasing interest in the use of new technologies, such as MRI, to measure body composition in patients with cancer. MRI uses magnetic fields and radiofrequency waves to measure the presence of hydrogens that are part of water and fat in the human body (96). Because the hydrogens in water act differently from the hydrogens in fat when excited by radiofrequency waves, a high contrast between fat and water can be achieved. This feature makes MRI an ideal modality to measure body composition (97) (Figure 3).

Figure 3.

MRI-based Dixon segmentation for body composition. Native Dixon “in-phase” and “opposed-phase” imaging is performed on a patient from the skull vertex to the thighs. In-phase and opposed-phased imaging is used to generate water maps where soft tissues containing water are bright and fat is black. Conversely, fat is bright and soft tissues are black on the water map, allowing for easy distinction of muscle and fat in body composition analyses. MRI = magnetic resonance imaging.

In contrast to other modalities such as CT or DEXA, a patient must undergo screening for medical devices that may be incompatible with magnetic fields, such as pacemakers, tissue expanders, or intrathecal pain pump, prior to the MRI (98). Once the screening is completed, the patient enters the room and lays down on the scanner that can be one of several strengths of magnetic fields such as 0.55 Tesla (T), 1.5 T, 3 T, and 7 T. Generally, stronger magnetic fields offer higher quality examinations, although this may not be necessary for body composition analysis (99). Additional magnetic coils may be placed on the patient in the anatomic region of interest (eg, abdomen and pelvis) to increase the signal and image quality.

Also as opposed to other imaging modalities, MRI uses multiple types of sequences in succession to image the region of interest. The advantage of using multiple sequences is that different types of tissue contrast can be achieved; thus, MRI is able to highlight the presence of abnormalities on the patient’s scan. One frequently used sequence that was designed to image water and fat is chemical shift (ie, Dixon) imaging. Dixon imaging generates quantitative maps for fat and water fraction in tissues with relatively high spatial resolution, comparable to CT. This technique may have an additive advantage over CT, as water and fat can be unequivocally measured, whereas CT relies on the attenuation of the tissue that is a function of multiple components of matter besides water and fat (100,101). As described above with the use of CT in measuring myosteatosis, such an advantage may be evident in the distinction of intermuscular and intramuscular fat deposition, as intramuscular fat can be associated with insulin resistance, inflammation, and visceral adiposity (102).

Another important application of MRI that uses Dixon imaging is the quantification of hepatic steatosis. The principle and the analysis of hepatic fat quantification is the same as described above. The measurement of hepatic steatosis is important, as it is associated with metabolic syndrome and cardiovascular disease (103-105). This is important for the cancer patient population, as hepatic steatosis could result from drug injury, may be associated with increased risk of postoperative morbidity and mortality, and even be associated with reduced efficacy of immunotherapy (106). Overall, multiple studies indicate that MRI has an overall advantage over CT in the quantification of hepatic steatosis and that CT could play a role in the incidental identification of hepatic steatosis (107,108).

Despite these advantages, MRI has limitations that preclude its use as an “imaging workhorse” in body composition assessment. As noted above, not all patients can receive an MRI scan if they have certain medical devices or metallic objects in critical locations (98). In addition, MRI scans are more expensive than most imaging modalities (109,110) and generally take longer than CT scans, limiting use particularly in lower-resourced settings. As opposed to CT, which is an opportunistic method commonly used for standard patient care, MRI may be more commonly used for certain cancers such as brain and spinal cord tumors. However, the lack of ionizing radiation makes MRI an attractive modality in imaging specific patient populations who may be at increased risk from the effects of radiation such as children or patients with syndromes such as Li-Fraumeni (111).

Positron emission tomography

PET is a noninvasive procedure that involves the intravenous injection of a radiopharmaceutical followed by measurement of its binding and uptake within tissues. The most widely used radiopharmaceutical in the oncologic field is 2-[¹⁸F]fluoro-2-deoxy-D-glucose (FDG), which measures glucose uptake, a surrogate marker for glycolysis that is a well-characterized hallmark of cancer cells (112).

Clinically, FDG-PET is a sensitive imaging modality used to detect and stage cancer and to monitor the effectiveness of treatment (113). Because the radioactivity emitted by the radiopharmaceutical results in high signal and sensitivity, PET can typically identify earlier stage disease compared with CT because PET identifies metabolic changes that may be present prior to structural changes that result from a disease. FDG-PET, in contrast to other imaging modalities, can successfully stratify patient outcomes by quantifying tumor metabolism; higher glucose uptake in tumors positively correlates with more aggressive disease and poorer prognostic outcomes (114) (Figure 4, A).

Figure 4.

Merging of PET and CT or MRI to study interactions of tumor metabolism and body composition. Male aged 50 years with metastatic poorly differentiated lung adenocarcinoma. A) FDG-PET imaging with co-registered simultaneously performed anatomic CT and merged (ie, PET-CT) modalities demonstrates numerous FDG-avid lytic metastatic osseous lesions throughout the axial skeleton (arrows). The primary adenocarcinoma within the left lower lobe is highlighted (arrowhead). B) FDG-PET/MRI imaging performed on the same patient shows the same distribution of disease. Body composition analyses can be performed on the CT or MRI datasets along with FDG uptake in tumor, fat, and muscle on the corresponding PET imaging. CT = computed tomography; FDG = 2-[¹⁸F]fluoro-2-deoxy-D-glucose; MRI = magnetic resonance imaging; PET = positron emission tomography; PET/CT = hybrid modality PET and CT; PET/MRI = hybrid modality PET and MRI.

In contrast to CT, FDG-PET is capable of identifying and distinguishing white and brown adipose tissue (BAT). Studies in humans have shown that FDG uptake is increased in BAT (115-117). This is informative for studies evaluating response to cancer treatment and survival outcomes, with growing evidence that individuals with cancer and higher BAT volume have poorer survival (118).

FDG uptake can also help distinguish VAT, which has higher glucose uptake (119) and is positively correlated with greater cardiometabolic risk (120), from SAT. There is growing evidence that FDG-PET can provide evidence toward a greater understanding of differences in biological activity between individuals who are metabolically healthy obese (MHO), metabolically abnormal obese (MAO), and metabolically healthy lean. MAO individuals are obese (defined as BMI > 30 kg/m²) and have metabolic syndrome (high blood pressure, high cholesterol, and insulin resistance), therefore MAO individuals also have a greater risk of cardiovascular disease. A critical gap in our understanding of obesity (based on BMI) and disease risk and prognosis is the existence of a subset of obese individuals without metabolic syndrome, MHO, who still have a higher risk of cardiovascular disease despite the lack of metabolic syndrome (121). Recent studies using FDG-PET have shown that MHO and MAO individuals have abnormal glucose metabolism (lower FDG uptake) in VAT, uncovering a potential mechanism for the aforementioned association (119). Some studies among patients with cancer provide evidence that increased VAT activity, defined as the ratio of the maximum standardized uptake value of VAT to the maximum standardized uptake value of SAT (the V/S ratio) predict lymph node metastasis in cancers including breast and thyroid (122,123). These findings highlight the critical need to assess glucose metabolism in specific types of adipose tissue rather than, or in addition to, overall adipose tissue.

FDG-PET can also evaluate skeletal muscle biology, which is responsible for uptake of the majority of glucose after a meal and thereby can provide a measure of insulin resistance. FDG-PET offers a convenient alternative to more traditional methods of insulin resistance measurement such as the hyperinsulinemic-euglycemic clamp (124). Many studies have identified an association between metabolic syndrome or insulin resistance and poor prognosis among patients with cancer, and recent studies have begun to capitalize on the utility of FDG-PET for these purposes (125-128).

PET can be combined with other imaging modalities, such as CT or MRI, to gather more information with which to predict cancer outcomes. In fact, in the clinical setting, PET scans are typically performed simultaneously with CT to correlate radiotracer activity with anatomy and to correct the PET signal for the attenuation of the tissue. The use of body composition measurements from simultaneous PET and CT are more scarce, but a recent study among patients with diffuse large B-cell lymphoma, for example, identified that the combination of longitudinal changes of visceral fat quantity and FDG tumor uptake over the course of treatment resulted in more robust stratification of females (129). Such characterization of survival outcomes relies on FDG-PET (for glucose metabolism) and CT (for visceral fat quantification) and may help explain the biological mechanisms of sex differences in survival and response to treatment. PET-MRI, however, is not as widespread as PET-CT but is increasing in utilization in the imaging of patients with cancer (130,131). As described above, MRI uses nonionizing radiation, so PET-MRI is therefore uniquely suited to evaluate whole-body imaging and repeated imaging, which allows for the evaluation of longitudinal changes, such as with whole-body imaging of tissue-specific insulin sensitivity and the quantification of glucose metabolic rate, lipid content, and perfusion in BAT (132-134) (Figure 4, B).

The limitations of PET include exposure of the patient to ionizing radiation and may therefore be problematic in the design of studies in healthy populations. FDG-PET does require preparation and advanced planning; all individuals must fast for a minimum of 4 hours prior, and those who are pregnant, breastfeeding, or have diabetes (type I or type II) may require further consideration or preparation (113). However, some of these limitations may be mitigated by advanced long axis field of view PET scanners that reduce dose and scan time with improved image quality (135).

Future directions

In summary, multiple methods and technologies can be used to measure body composition. Anthropomorphic assessments of body composition, such as BMI, are extremely popular throughout the clinical field given their relative ease of use and immediate clinical impact. However, as described above, limitations exist in the use of these methods. This is an ideal opportunity for imaging-based assessment of body composition, as spatially resolved imaging allows for identification and quantification of fat, muscle, bone, and even visceral organs that anthropomorphic assessments cannot provide.

Unfortunately, the abundant success of imaging in the care of patients with cancer has led to a major problem (and a bottleneck) in body composition analysis workflows—how are muscle and fat quantified in imaging datasets? Historically, use of imaging to evaluate body composition has been limited by the need for manual identification and segmentation of images (93). As an example, if visceral fat, subcutaneous fat, and muscle measurements are needed from a conventional abdominal CT with 200 slices, requires 600 segmentations per patient are needed. This not only is logistically improbable but requires time-intensive processes operated by personnel with training in anatomy and software applications. This has resulted in a plethora of software platforms that manually estimate body composition using a single cross-sectional area of the abdomen (136). However, advancements in imaging software using artificial intelligence have enabled automated segmentation and analysis of routine clinical images, providing real-time measures of body composition (83,88). This, in turn, has led to the generation of 3D segmentation and multilevel body composition analyses (Figure 2) that would have otherwise been impractical with manual segmentation (137). As these algorithms continue to improve along with ease of use for the scientific community, this will undoubtedly advance our understanding of the role of body composition in treatment response, toxicities, and outcomes in patients with cancer. this 

Despite its clinical importance, variation in the methods and definitions available for determining body composition has been a barrier to adopting these measurements within research and supportive oncology clinical care. One of the major limitations is that there are currently no standardized thresholds for classification of abnormal muscle mass or adiposity (81). For example, despite the widespread use of the L3 vertebral level for sarcopenia estimation (138), recent consensus guidelines state that such techniques require more testing for validity, reliability, and accuracy (70). As new automated software platforms allow for 3D segmentation, this will likely open new possibilities in how body composition is analyzed and reported. This idea is supported by reports stating that alternative vertebral or anatomic landmarks may also give important information on outcomes (139,140). Some studies have included sex- and/or BMI-specific thresholds (88), whereas others have defined thresholds relative to the outcome of interest (141). These thresholds are inherently study specific, varying among different patient populations. Broadly applicable thresholds will be difficult to establish without the availability of representative CT images from the general population.

It is becoming increasingly clear that one modality may not be able to provide all of the information needed for specific body composition measurements. For example, DEXA for body composition is relatively inexpensive but is not measured clinically, whereas CT provides outstanding anatomic detail but no measurement of tissue function. Conversely, PET provides unique data on tissue function and metabolism but poor anatomic detail. Although MRI is traditionally considered to provide primarily anatomic detail, new advancements in MRI contrast agents and spectroscopy allow for the real-time dynamic assessment of cellular metabolism that other modalities cannot provide (142,143). Therefore, the development of hybrid imaging modalities such as PET-CT and PET-MRI allow the combination of traditional measurements of body composition such as fat and muscle mass as well as real-time tissue nutrient uptake and metabolism.

Of note, other quantitative techniques have been used to measure body composition that have the ability to complement imaging methods. For example, D₃-creatine (D₃-Cr) dilution involves the oral administration of deuterated creatinine that is absorbed and stored in muscle sarcomeres. In metabolically active muscle, D₃-Cr is metabolized to D₃-creatinine that is ultimately excreted in the urine. Measurement of urine D₃-Cr is thus directly associated with whole-body skeletal muscle creatine enrichment and, therefore, functional muscle reserve (144). Overall, studies suggest that D₃-Cr dilution has better accuracy than DEXA in predicting adverse outcomes related to sarcopenia (144-146). Although there is some evidence that D₃-Cr and MRI provide similar results (146), little data exist to determine if D₃-Cr functional muscle measurements can synergize with CT and MRI muscle mass measurements. Much more work needs to be developed in the field as these different measurements are combined in prognostic models.

Data availability

No new data were generated or analyzed in support of this research.

Author contributions

Conceptualization: JEI. Project Administration: JEI. Supervision: JEI. Methodology: UAS, TJB, RB, CMD-C, KAG, EH, FK, AEL. Writing—original draft: UAS, TJB, RB, CMD-C, KAG, EH, FK, AEL, JEI. Writing—review & editing: UAS, TJB, RB, CMD-C, KAG, EH, FK, AEL, JEI.

Funding

Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under Award Number, K12 CA184746 (UAS), K12 CA001727 (RB), T32 CA193193 (FK), K99/R00 CA218869 (JEI), R21 CA242221 (JEI) and MSK Cancer Center Core Grant P30 CA008748 (UAS). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or other funders. UAS is also supported by the American Society of Hematology CRTI award, the International Myeloma Society Career Development Award, the Paula and Rodger Riney Foundation, the HealthTree Foundation, and the Allen Foundation.

Conflicts of interest

UA Shah reports grants from Parker Institute for Cancer Immunotherapy at MSK, other research support from Celgene/BMS and Janssen to the institution, personal fees from ACCC, MashUpMD, Janssen Biotech, Sanofi, BMS, MJH LifeSciences, Intellisphere, Phillips Gilmore Oncology Communications and RedMedEd, outside the submitted work.

AE Lohmann reports financial support kind from Epic Sciences and honoraria from La Roche Posay and Novartis outside the submitted work.

The rest of the authors have no conflicts of interest to disclose.

Acknowledgements

All authors were a part of and gratefully acknowledge the TREC Training Workshop R25 CA203650 (PI: Melinda Irwin).

Role of the funder: The funder did not play a role in the design of the study; the collection, analysis, and interpretation of the data; the writing of the manuscript; and the decision to submit the manuscript for publication.