Body Composition Core

This core provides body composition and physical performance measurements for the Obesity Research Center.

Body composition measurements include anthropometry, bioimpedance analysis (BIA), dual energy x-ray absorptiometry (DXA), underwater weighing, body volume by Bod Pod, magnetic resonance imaging, whole-body counting for 40K, total body water, and sulfate and bromide dilution spaces. Physical performance measurements made in the Core laboratory and include handgrip strength, respiratory muscle strength, quadriceps strength, adductor pollicis function, and aerobic fitness evaluation.

During the late 1960’s and early 1970’s there were important unanswered questions related to the efficacy of very low calorie diets in promoting fat loss. Seeking a method to quantify the small changes in total body fat needed to examine these diets, Drs. Mei Yang and Theodore B. VanItallie developed the resources needed to perform energy-nitrogen balance studies at St. Luke’s-Roosevelt Hospital (1977).

The Body Composition Core, located in the basement of the Plant Building at St. Luke’s Hospital, was first developed in 1967 under Atomic Energy Commission and John A. Hartford Foundation support. The Four-pi whole body liquid scintillation counter (HUMCO II) was installed with Hartford Foundation support. Dr. Richard Pierson headed the newly created facility.

Thermogenesis measurements, food composition analyses, and urine and fecal energy-nitrogen content were all evaluated in VanItallie and Yang’s now classic experiments designed to examine body composition changes with dieting (1976). The present facilities are an outgrowth of these initial efforts.

Subsequent studies of thermogenesis, food intake, and physical activity in obese patients required further expansion of the Body Composition, Energy Expenditure, and Physical Performance Laboratories (Pi-Sunyer, 1986).
FIGURE

This growth provided the setting to critically evaluate such new body composition methodologies as total body electrical conductivity (TOBEC)(Van Itallie et al. 1985), BIA (Segal et al. 1985), dual photon/x-ray absorptiometry (DPA/DXA)(Heymsfield et al. 1989; 1990), and inelastic neutron scattering (INS)(Kehayias, 1990). A multicenter NIH program project grant was awarded to Center investigators in 1990. Headed by RN Pierson Jr., the PPG tied together unparalleled laboratory resources that include those already mentioned at SLRHC and those at Brookhaven National Laboratory (BNL) in Long Island.

Scientists at the Center developed multicomponent body composition methods that exploit these unique body composition methods (Heymsfield et al. 1991). Center scientists recently extended their efforts to imaging and nuclear magnetic resonance spectroscopy methods in collaboration with colleagues at Columbia University and in the Department of Radiology at St. Luke’s-Roosevelt Hospital. Promising studies are underway to detect early muscle edema and loss in cell mass using proton, sodium, and phosphorus spectroscopy. Finally, investigators in the Core (Dr. Heymsfield) were recently funded on two research projects related to aging, one involving skeletal muscle mass measurement and the other to evaluate a novel method of measuring body volume and body composition.

The Core consists of several consolidated research laboratories and groups working on related projects in close proximity. Dr. Pierson is head of the Body Composition Unit and Dr. Heymsfield is the Human Body Composition Core’s Director. Dr. Heymsfield also supervises physical performance testing and, in association with Dr. Pi-Sunyer, coordinates studies of energy metabolism in the respiratory chamber.

Dr. Heymsfield is Professor of Medicine and Deputy Director of the Obesity Research Center. Dr. Richard N. Pierson, Jr. is Professor of Clinical Medicine and a well recognized authority in the area of human body composition research. He pioneered whole-body counting and isotope dilution methods (Pierson et al. 1982) and more recently was among the first investigators to examine DPA and DXA (Wang et al. 1989). He is also an expert in patient recruitment, notably minority populations. Mr. Jack Wang is a Research Associate and a well known scientist in the field of body composition research. He has made major contributions over the past two decades in the areas of isotope dilution methods, whole-body counting, and bioimpedance analysis. Mr. Chris Nuñez is a doctoral student at Columbia University and holds an NIA Minority Pre-Doctoral Training Award.

RECENT AND ONGOING RESEARCH

Obesity Research Center investigators are actively involved in developing and implementing new human body composition methodologies. The following describe several representative areas of research and development that show the great variety and depth of investigations carried out by Core investigators.

Body composition concepts and methods that can be applied in elderly:
Development and Testing of Body Composition Models.
The study of body composition until recently lacked a formal organizational structure. A major outgrowth of Core efforts was the development of new concepts related to the organization of body composition research. Specifically, Dr. ZiMian Wang and Drs. Pierson and Heymsfield suggested that the study of body composition can be viewed as three interconnected areas, the five-level model, methodology, and biological effects on body composition (Wang et al. 1992). In this report the investigators suggested specific definitions for body composition components and described how these components can be organized into five levels of increasing complexity: atomic, molecular, cellular, tissue-system, and whole-body . In a follow-up report published this month (Wang et al. 1994) they developed a systematic approach to the mathematical description of body composition levels and then showed that simultaneous equations can be written that solve for each body composition component. These two studies provide a foundation for the study of body composition and in particular create a framework within which to organize body composition assessment methods.

The field of body composition research has, until recently, relied almost entirely on the "two-compartment model" to estimate total body fat and fat-free body mass (FFM). The investigators (Heymsfield & Lichtman 1990) and others (Lohman 1981) were critical of the three main methods based on total body water, potassium, and density that were traditionally used to solve for these two components. This is because the "models" or constants used in these methods were developed mainly in young white men and had not been adequately validated for use in women, non-whites groups, or the elderly. A major effort over the past few years has therefore been to develop or validate body composition methods that are applicable across gender, ethnic, and age groups (Heymsfield et al. 1990).

The first efforts by Core investigators involved the development of fundamental body composition models at the atomic level of body composition (Wang et al. 1993). The strength of "atomic-level" methods designed to estimate chemical components is that they are assumed to be valid in all adults regardless of ethnicity and age. This is because, as far as is known, elemental proportions of chemical components (e.g., N/protein=0.161, etc.) are extremely stable in the population as a whole. In collaboration with BNL scientists (Dilmanian, 1990) and with Dr. Joseph Kehayias of the USDA Center at Tufts University in Boston, we developed two substantially improved atomic-level methods. The first is designed to estimate six major chemical components in vivo from 11 measured and calculated elements from neutron activation analysis/whole-body counting (Heymsfield et al. 1991). This new series of simultaneous equations allows us to chemically reconstruct the human body much in the same way that was possible only with cadavers in previous studies. The methods developed also allow for a probing analysis of two-compartment chemical methods that rely on assumed values of FFM potassium, water, and density.

The second method was developed specifically to estimate total body fat from total body carbon and other measured elements (Kehayias 1990). The total body carbon method of estimating fat relies on the inelastic scattering neutron activation technique developed at BNL by Dr. Kehayias (Kehayias 1991). These two atomic-level approaches for estimating chemical components provided us with an unprecedented opportunity to study human body composition in a manner never before possible. It has been possible to study 660 human subjects, 605 of the normal volunteers, in a series of cross-sectional and longitudinal studies, with the support of a PPG from NIH (DK 42618, 1990-1995) directed from this laboratory (RN Pierson Jr. PI).

In the next phase of the Core group’s efforts they recognized that "technology transfer" of these complex and costly neutron activation methods could not take place without the orderly development and validation of "surrogates" as less expensive and more practical alternatives. The opportunity to develop alternatives to the neutron activation approaches came during the late nineteen eighties. The DPA method for studying bone mineral was first being installed in our laboratory in 1988 and this was later followed by the improved DXA method. The investigators recognized with the introduction of this technology the possible alternative role it might play to delayed gamma neutron activation (DGNA) for estimating total body Ca and bone mineral. They subsequently were able to show excellent agreement between bone mineral mass estimated by DPA and DXA to total body calcium estimated by DGNA (Heymsfield et al. 1989; 1990). Others confirmed these studies in human and animal cadavers.

With a validated measure of bone mineral mass, Core investigators were then able to accomplish a long sought after goal of developing a "four-compartment" chemical model in which the human body is divided into fat, water, mineral, and protein/glycogen. The method involves three measurements, body density by underwater weighing, bone mineral mass by DXA, and total body water by tritium or deuterium dilution. They were able to show that this four-compartment method, which can be developed in many laboratories around the world, compares very favorably to the six-compartment neutron activation method (Heymsfield & Locki 1991). Additionally, the four-compartment method allows computation of density and hydration of FFM, and, like the more complex neutron activation method, is assumed valid across all adults. This is because no assumptions are made in the four compartment method that are known to be influenced by gender, ethnicity, or disease.

As DXA methodology improved, it became clear that it also might provide a means of estimating total body fat that is independent of the traditional two-compartment method assumptions. In a series of studies, some of which are still being carried out, the core scientists were able to demonstrate the accuracy of DPA and DXA fat estimates compared to those provided by neutron activation analysis (Pierson et al. 1993). Their most recent work in this area is examining the validity of DXA soft tissue fat estimates in the presence of changes in tissue hydration. This same work, based on both in vitro and in vivo studies, is examining the DXA soft tissue model in relation to aging and ethnicity. Initial findings indicate that DXA fat estimates are minimally influenced by ethnicity, aging, and small changes in tissue hydration.

When the research in this area was started by the group in 1990, the field of body composition research relied on "two-compartment" chemical methods that were based on thirty year old models with known limitations. Today, due to their research described above and that of many other groups around the world, extremely powerful body composition methods are available that can reliably estimate total body fat and other major components at the molecular level of body composition.

Evaluation of Bioimpedance (BIA) Methodology.
Dr. Karen Segal was one of the pioneers in evaluating BIA methodology in our Core Laboratory when it was introduced approximately nine years ago (Segal 1985). More recently Dr. Segal and others in the laboratory began and then followed up on studies of a second generation BIA instrument that provides body composition estimates at multiple frequencies. The potential exists to use this instrument in estimating extracellular and intracellular water and several projects in this area are underway. We will also describe in a later section (6.2) new potential methods of measuring muscle mass with BIA. Core investigators contributed to the planning and lectures at the recent NIH Technology Assessment Conference on BIA. Dr. Heymsfield will be a co-editor of the published proceedings that will appear as a supplement to the American Journal of Clinical Nutrition.

Testing Dual Photon Techniques.
Early studies of DPA estimates of bone mineral were carried out by Drs. Heymsfield (1990) and Pierson and Mr. Wang as described in the previous section (1989). The investigators also demonstrated the applicability of DPA in estimating appendicular skeletal muscle mass and total body fat (Heymsfield & Waki, 1991). The group is now critically evaluating the newer dual photon systems based on dual-energy X-ray sources (DXA).

A major contribution made by laboratory investigators was the recognition of thickness artifacts in body composition estimates by DXA. Subsequent software corrections by the manufacturer (Lunar Radiation) substantially improved body composition measurements by DXA in obese and very lean subjects. Center investigators are also critically examining the two-compartment DXA model for quantifying fat in order to assess the importance of hydration changes in altering underlying DXA model assumptions. This research is being carried out in collaboration with BNL scientists. Finally, the research group as part of the current PPG has carried out a major study of DXA instrument validity and reliability in collaboration with USDA/Tufts scientists. These studies are described in later sections related to DXA validity/reliability in the context of Health ABC.

Magnetic Resonance Imaging/Nuclear Magnetic Resonance Spectroscopy (NMRS).
St. Luke’s-Roosevelt Hospital and Columbia University College of Physicians and Surgeons have General Electric Signa MRI scanners with 1 meter bore-1.5 Tesla magnets and version 5.0 software. The systems provide magnetic field homogeneity to facilitate high resolution imaging. Dr. Frager supervises the St. Luke’s system, which is housed near the body composition laboratories and immediately adjacent to proposed Health ABC facilities. We have been working closely with the Radiology Department and Dr. Frager for the past year developing MRI methods of measuring total body skeletal muscle mass and visceral adipose tissue. The study of methods for estimating skeletal muscle mass represents a major research direction of the lab’s investigators.

The MRI/NMRS laboratory is about 10 minutes by car from St. Luke’s Hospital. The MRI scanner at Columbia Presbyterian Hospital is equipped to carry out both 1H and 31P spectroscopy using commercially developed hardware and software provided by GE. Dr. Katz and others we currently collaborate with are now completing installation of a 5 Tesla whole-body magnet at the Neurological Institute at Columbia.

Skeletal Muscle Mass.
Skeletal muscle is one of the most difficult components to quantify. Investigators at the Center, including Drs. Pierson, Heymsfield, Frager (St. Luke’s), Katz (Columbia), Matthews (Cornell), and Weir (Columbia) are in the process of developing and carrying out protocols to evaluate the major methods of quantifying skeletal muscle mass. Among the methods under study are whole-body muscle measured by MRI, CT, neutron activation methods, 24 hour urinary creatinine and 3-methylhistidine excretion, anthropometry, and appendicular muscle measured by DXA.

Each method measures a specific skeletal muscle component. For example, creatinine excretion is a measure of muscle cell mass, anthropometry provides a measure of anatomic skeletal muscle, and so on. These distinctions have not been widely appreciated nor have the relationships between different methods been fully evaluated. Some examples of ongoing research are the following:

An important question arises whether or not methods reflect the same amount or rate of change in muscle with aging. This question was examined by matching young and old women by weight (±2 kg) and height (±3 cm). Matching removes some of the concerns related to the effects of body size on muscle mass and measurement accuracy. All of the women were healthy and physically active. The young and old groups were ages(x ± SD) 29.9 ± 4.4 and 74.2 ± 6.7 years, respectively. Three indices of muscle were examined, total body K, appendicular muscle from DXA, and anthropometric limb muscle areas. Old women had 17% less total body potassium compared to young women (p<0.001).

In contrast, appendicular skeletal muscle by DXA was reduced by only 11% in the old women (p=0.007). An even smaller difference in muscle between the two age groups was detected using anthropometric methods; arm muscle area, 4% (p=NS); calf muscle area, 3% (p=NS); and thigh muscle area, 6% (p=0.005).

This example shows that the relative loss in muscle with age in women depends on which method is used to assess muscle mass. The basis for these discrepancies are being explored.

In another study, investigators at the Center measured total body skeletal muscle mass in 37 healthy men and men with AIDS using multiple slice cross-sectional CT. An example of the findings is that appendicular muscle mass measured by DXA correlated highly (r=0.93, p<0.001) with total body muscle mass. The ratio of appendicular muscle to total muscle was about 0.8 and was relatively constant between subjects. DXA may therefore be a useful method of measuring both appendicular and total muscle mass (Z Wang, manuscript submitted).

The final example involves the TBK/TBN method of evaluating skeletal muscle mass and its relationship to total body protein. The TBK/TBN method was used in classic studies that demonstrated muscle loss with aging. There are, however, two concerns with these results. First, on average, subjects of all ages studied by the TBK/TBN method had less muscle than observed in earlier cadaver studies. Second, some very old subjects studied using the TBK/TBN method had negative values for total body muscle mass. The TBK/TBN method was examined in the cohort of healthy men and men with AIDS described earlier. Subjects completed multislice CT, whole-body 40K counting for TBK, and prompt gamma-neutron activation analysis for TBN. The results demonstrated that the TBK/TBN method substantially underestimates (~25%; p<0.001) adipose tissue-free skeletal muscle mass by CT. The exact cause of the underestimate of muscle by the TBK/TBN method is unknown, but it is likely that the assumed TBK/TBN ratios of muscle and non-muscle lean differ from those assumed and vary between subjects. Clearly, more research on the underlying assumptions of this method is needed.

African-American compared to white subjects.
Core investigators hypothesized that African-American women have greater bone mineral mass than do comparable white women and also that they have a greater density of FFM than white women. This increase in FFM density would in turn cause a bias in fat estimates in African-American women because the underwater weighing method/model assumes a constant FFM density of 1.100 g/cc. This hypothesis was examined in 29 African-American and white women matched on age, weight, and height (Ortiz et al, 1992). This study also explored the validity of other classic two-compartment chemical model "constants", TBW/FFM and TBK/FFM, when applied across ethnic groups. Total FFM, hydration, density, potassium, and bone mineral content of FFM were estimated by the four-compartment method described above along with 40K whole body counting. DXA was used to estimate skeletal characteristics, including femur, tibia, humerus, and spine lengths in the women. Results overall demonstrated that matched African-American and white women had a non-significant difference in total body fat and FFM. In contrast, African-American women had significantly greater bone mineral mass (p<0.001) and density (p<0.05), appendicular bone lengths (all p<0.001), TBK (p<0.05), and FFM density (p<0.05) compared to the white women. There were no significant differences observed between the two groups of women in FFM hydration. In a follow-up study of 19 African-American and white women (Gasperino et al. 1994) the group showed that bone mineral mass and density were significantly greater in African-American than in white women between the ages of 20 and 80 yrs and the post-menopausal decline in bone mass was similar between ethnic groups in this cross-sectional cohort.

As there was relatively little modern literature on comparison of body composition in African-American and white men, the investigators expanded their investigation by carrying out a similar study in men as described above for women. Adult African-American and white men were matched on age, height, and weight (Gerace et al. 1994). African-American men had significantly greater total bone mineral mass (p<0.002), bone density (borderline, p<0.07), and appendicular bone lengths (all p<0.001) compared to the matched white men. However, hydration of FFM was significantly greater in African-American men (p<0.05), and this offset the greater bone mineral mass with the result that FFM density was similar between the two groups. There was no significant difference between the groups in TBK. These findings to a large extent are concordant in the men and women: adult African-American subjects differ significantly in multiple skeletal characteristics when compared to white subjects of similar gender, age, height, and weight. There also appear to be gender differences in these effects, at least as suggested by the results in the Centerís two studies of relatively small numbers of subjects.

In sum, studies in women and men clearly demonstrate the existence of cross-ethnic differences in body composition and thus body composition methods cannot necessarily rely on models developed without consideration for ethnicity. Many other ethnicity-related body composition studies have been carried out at the Center over the past several years.