Assessment of fat and lean mass by quantitative magnetic resonance

a future technology of body composition research?

Abstract

Purpose of review 

For the assessment of energy balance or monitoring of therapeutic interventions, there is a need for noninvasive and highly precise methods of body composition analysis that are able to accurately measure small changes in fat and fat-free mass (FFM).

Recent findings 

The use of quantitative magnetic resonance (QMR) for measurement of body composition has long been established in animal studies. There are, however, only a few human studies that examine the validity of this method. These studies have consistently shown a high precision of QMR and only a small underestimation of fat mass by QMR when compared with a 4-compartment model as a reference. An underestimation of fat mass by QMR is also supported by the comparison between measured energy balance (as a difference between energy intake and energy expenditure) and energy balance predicted from changes in fat mass and FFM. Fewer calories were lost and gained as fat mass compared with the value expected from measured energy balance.

Summary 

Current evidence in healthy humans has shown that QMR is a valid and precise method for noninvasive measurement of body composition. Contrary to standard reference methods, such as densitometry and dual X-ray absorptiometry, QMR results are independent of FFM hydration. However, despite a high accuracy and a low minimal detectable change, underestimation of fat mass by QMR is possible and limits the use of this method for quantification of energy balance.

INTRODUCTION

Importance and challenge of measuring small changes in body composition

Chronic disturbances of energy balance are discovered to be a major cause of lifestyle diseases, and therefore it is of great interest to measure energy balance during intervention studies either as a confounder or as the target parameter. Because the accurate assessment of energy intake and total energy expenditure over longer periods is impractical, cumbersome and expensive, accurate measurement of changes in body composition assuming constants for energy equivalents of fat mass and fat-free mass (FFM) is an attractive alternative ^[1].

In addition, the relative contribution of fat and lean mass during weight loss or weight gain is required to evaluate intervention strategies aimed to improve lean mass maintenance during weight loss or accretion during the treatment of sarcopenia and wasting.

However, limitations in the precision of deuterium dilution, densitometry or dual X-ray absorptiometry restrict the minimal detectable change of these 2-compartment methods ^[2]. In addition, weight changes are a nonsteady state of body composition that leads to transient changes in FFM-composition (e.g. nitrogen, glycogen and water content) that violates the assumptions of 2-compartment methods and leads to a method-inherent bias ^[2]. The 4-compartment (4C)-model avoids the assumption of a fixed composition of FFM by measuring its constituents (water and minerals), and is therefore more accurate during unstable conditions of weight changes that do not meet these assumptions ^[2]. However, the method combination makes the 4C-model elaborate and expensive as well. There is a need for both accurate and precise alternative methods of body composition analysis to overcome these drawbacks.

Box 1: no caption available

Characteristics of quantitative magnetic resonance technology

Quantitative magnetic resonance (QMR) instruments are based on nuclear magnetic resonance (NMR) and are produced for different sizes ranging from tissue samples more than 0.3 g to small animals, infants and children up to adults and big animals less than 250 kg (EchoMRI LLC, Houston, Texas, USA, see Fig. 1 for illustration). Contrary to MRI, QMR requires only a low magnetic field (67 Gauss = 0.0067 Tesla) that can be obtained without complex equipment that entails high maintenance costs. Although QMR has already become a standard of in-vivo body composition analysis in animals, only a few instruments are used worldwide in humans. The output of QMR is a result on fat mass, lean mass (without solid components that are mainly located in bone, [3]) as well as total and ‘free’ body water. Because FFM resembles all body components except for lipids, FFM_QMR can be calculated as body weight − FM_QMR. The QMR instrument is calibrated with 45 kg canola oil (Wesson, ConAgra Foods Inc., USA). According to the manufacturer, 1 g canola oil at 37°C resembles 1 g of human fat mass.

FIGURE 1: EchoMRI LLC (Houston, Texas, USA) with a typical position of an individual before being moved inside the instrument.

QMR technology is based on modification of spin-patterns of protons in a magnetic field by radiofrequency pulses with each scan producing a series of NMR responses. The sequence comprises alternating periodic Carr−Purcell−Meiboom−Gill parts and pauses of different durations that are designed to capture all relevant characteristic (relaxation) time scales of the NMR responses (transverse and longitudinal relaxation) typical for fat, lean and free water ^[4]. Because of a higher proton density, fat has a higher amplitude and relaxation rate compared with lean tissue. The QMR signal for the total body combines signals of fat, lean and free water that can be separated because of differences in relaxation times between these components and deriving linear regression equations from calibrations to phantoms containing fat (canola oil), lean mass (lean pork) and free water (tap water). The algorithm for optimizing these regression formulas employs partial least squares optimization combined with principal component analysis for high-dimensional regressions ^[4].

Precision and validity of quantitative magnetic resonance

Measurements of fat mass by QMR have a low standard deviation (0.133−0.221 kg) and coefficient of variation (0.44–0.69%, [4]). QMR is thus more precise than all conventional body composition methods accepted as a reference standard and is able to detect more than 250 g changes in fat mass or more than 1 kg weight change ^[4,5]. The lack of radiation exposure is an additional advantage for repeated measurements.

Only three groups so far have investigated the validity of QMR vs. a 4C model as a gold standard reference in cross-sectional studies ^[3–6]. Andres et al.^[3] found a sizeable overestimation of fat mass by QMR (97%) compared with a 4C-model in 10 infants and less overestimation (10%) in 57 children above 6 years. The QMR bias vs. 4C-model was, however, lower when compared with the bias of other reference methods (dilution, densitometry and dual X-ray absorptiometry) compared with the 4C-model. In a previous work, the same group also compared the accuracy of QMR with chemical cadaver analysis in 50 piglets and pigs ranging from 3 to 49.1 kg ^[7]. Results from half of the animals were used to calibrate the QMR-algorithm and the other half were used for cross-validation of the algorithm. The authors found an overestimation of fat mass in piglets, whereas in adult pigs fat mass was underestimated by QMR ^[7].

Consistent with the results in adult pigs, Napolitano et al.^[5] and Gallagher et al.^[4] also found an underestimation of fat mass by QMR compared with a 4C-model in groups of 34 and 30 adult humans (2–4 kg, [5] and 4.66 ± 0.62 kg in men vs. 0.68 ± 0.27 kg fat mass in women; P < 0.001 for sex difference, [4]).

After an upgrade of the QMR algorithm by the manufacturer, the former group of authors also compared changes in body composition during weight loss in 11 healthy obese individuals and 11 patients with coronary heart disease or heart failure by QMR and 4C-model and found a significant underestimation (1 kg) of the decrease in fat mass by QMR ^[6].

An underestimation of fat mass loss by QMR is also supported by the comparison between measured energy balance (as a difference between energy intake and energy expenditure) and energy balance predicted from changes in fat and FFM during a controlled weight cycle in healthy men ^[8▪▪]. Fewer calories were lost and subsequently regained as fat mass and FFM compared with the value expected from measured energy balance (derived from the deviation between energy intake in controlled underfeeding and overfeeding and total energy expenditure derived from resting energy expenditure measured by indirect calorimetry and physical activity energy expenditure estimated from accelerometry, Table 1) ^[8▪▪]. During 3 weeks of caloric restriction, individuals were losing 16.200 kcal less calories as fat mass and FFM than what was expected from the difference between energy intake and energy expenditure. 1.8 kg underestimation of the decrease in fat mass by QMR and a similar underestimation of fat mass regain during the weight regain phase would fully account for this discrepancy. Considering the possibility of errors in the assessment of energy intake and energy expenditure cannot explain this sizeable discrepancy of ∼800 kcal/d during caloric restriction and ∼1150 kcal/d during refeeding. For example, an underestimation of activity-related energy expenditure and a decrease in diet-induced thermogenesis during weight loss and respective opposite effects during refeeding are implausible to explain a 35% lower than expected total energy expenditure during caloric restriction and a 40% higher total energy expenditure during refeeding because this would mean that daily energy expenditure for physical activity during the weight loss phase must have been higher than during the weight gain phase and is absent during refeeding.

Table 1: Comparison of energy balance during caloric restriction and refeeding calculated from body composition by QMR and the difference between energy intake and energy expenditure (means ± SD)

Discussion of the quantitative magnetic resonance-bias

The finding of fat mass overestimation in infants and piglets ^[3,7] and underestimation in adult humans and pigs ^[4,5,7] suggests that the bias by QMR may depend on the water content of the FFM that is higher in young organisms than in adult organisms. However, the direction of the QMR-4C bias with weight loss and regain argues against FFM hydration as an explanation. The overestimation of fat mass in infants and underestimation in adults would suggest that a higher FFM hydration leads to fat mass overestimation by QMR, whereas the decrease in FFM hydration with weight loss and respective increase with regain were both associated with an underestimation of fat loss and regain. The reason for the discrepant bias in infants and adults could, therefore, be because of limitations of the 4C-model that depend on the accuracy of the dilution technique. A systematic bias in total body water measurement has a major impact on the fat mass result by the 4C-model, and thus any underestimation or overestimation of FFM hydration by the 4C-model directly influences the bias in fat mass between QMR and 4C-model.

Fat mass measured by QMR is likely not dependent on FFM hydration but can be slightly affected by the amount of ‘free water’ that is not bound to organic solids and could resemble, for example the water in the bladder. Different sizes of water bottles (250 g, 500 g or 1 kg) as phantoms measured with the individuals led to a size-independent overestimation of fat mass by 114 g ^[4]. Half a liter of water in the bladder could thus lead to a 10% artifact of 50 g higher fat mass. This systematic bias is very low and can easily be avoided by requesting voiding as a standard procedure before each examination.

Because the relaxation time of water bound to protein and solid organic compounds can be neglected on the timescale of QMR, total body water is calculated as the difference between measured total amount of protons and protons derived from fat mass using regression analysis ^[4]. Therefore, measurements of fat and water by QMR are not independent from each other with a bias in 1 compartment leading to an inverse bias in the other.

Despite the underestimation of changes in fat mass during weight loss and regain, QMR is able to reflect the physiologic variations in fat and lean mass by showing a disproportional higher decrease in lean mass at the beginning of weight loss and a respective higher increase at the beginning of weight regain (Fig. 2) ^[9]. This phenomenon is mainly because of initial mobilization and reconstitution of glycogen stores that leads to predominant changes in lean mass by the high water binding properties of glycogen ^[10,9].

FIGURE 2: Changes in body weight and body composition during the course of the study (days 1–7: overfeeding, days 8–29 caloric restriction; days 30–43 refeeding, means +/− SD). FM, fat mass; FFM, fat-free mass. Data from ^[9].

Contrary to all 2-compartment methods, QMR was shown to accurately assess differences in fat mass loss between healthy dieters and patients with heart failure when compared with results from a 4C-model ^[6]. The higher loss in lean mass in heart failure patients was explained by a higher loss in body water (measured by deuterium dilution) that is likely due to natriuresis and loss of edema. Although dual energy X-ray absorptiometry (DXA) is not considered as a 2-compartment method because it measures fat mass, bone mineral content and lean soft tissue, every pixel of the DXA image can only be resolved into two components, and therefore requires assumptions on the hydration of FFM. DXA was shown to overestimate lean mass compared with QMR in mice with chronic renal failure ^[11]. A higher fluid content of lean soft tissue in renal failure can lead to an overestimation of lean mass and respective underestimation of fat mass because differences in X-ray attenuation that determine differences between fat mass and lean soft tissue are largely determined by the mineral content of tissue water (e.g. 0.9% NaCl physiological saline is interpreted as lean, whereas pure water is interpreted as fat). The relative validity of QMR vs. nongold standard methods as a reference was also recently investigated by Bosaeus et al.^[12▪] who compared QMR results with air-displacement plethysmography (ADP) in normal weight and obese women. The authors found that QMR overestimated fat mass values at low average fat mass values and underestimated fat mass at higher values (R² = 0.27, P < 0.001) and deduced that QMR cannot assure individual accuracy. This conclusion may, however, be questioned because of the limited validity of fat mass estimates by densitometry that is because of similar densities of fat and water. An increase in FFM hydration in obesity because of high water content of FFM in adipose tissue is, therefore, known to cause an overestimation of fat mass and a respective underestimation of FFM by ADP.

FFM QMR can be calculated from the difference between body weight and fat mass QMR. Lean mass QMR (FFM without solid compartments mainly located in bone) is, however, a measured value that has not been validated probably because of uncertainty about the composition of this compartment or because of a lack of suitable reference method that is able to measure lean mass without bone. Because FFM QMR has been calibrated using lean pork as a reference standard, lean mass QMR most likely resembles muscle tissue. The validity of this measurement partly depends on the relationship between water molecules and muscle protein or glycogen and on the extent that this relationship is transferable to other lean tissues such as organs and extracellular fluid. Further studies are, therefore, required to understand the anatomical correlate as well as the validity of lean mass QMR.

CONCLUSION

QMR is able to accurately measure fat mass even when hydration of lean mass is altered. QMR is also able to detect small changes in fat mass in individuals and not only at the population level. The method, therefore, saves resources in intervention studies and by detection of early changes in body composition facilitates monitoring and optimizing therapeutic options. QMR should not be validated against 2-compartment methods in conditions that are accompanied by alterations in FFM hydration (e.g. obesity, weight loss, renal or heart failure). Accuracy of QMR to assess changes in fat mass and FFM with weight changes needs to be further investigated using chemical analysis in animal studies.

Acknowledgements

None.

Financial support and sponsorship

The studies by the authors were funded by a grant of the Germany Ministry of Education and Research (BMBF 0315681), and the German Research Foundation (DFG Bo 3296/1-1).

Conflicts of interest

There are no conflicts of interest.

REFERENCES AND RECOMMENDED READING

Papers of particular interest, published within the annual period of review, have been highlighted as:

• ▪ of special interest
• ▪▪ of outstanding interest