.1,71 i C/ui Nuir 1992:56: 19-28. Printed in USA. 1992 American Society for Clinical Nutrition 19

The five-level model: a new approach to organizingbody-composition research12Zi-ifian Wang, Richard N Pierson Jr. and Steven B Heyms/leld

ABSTRACT Body-composition research is a branch of hu-man biology that has three interconnecting areas: body-com-position levels and their organizational rules, measurementtechniques, and biological factors that influence body compo-sition. In the first area, which is inadequately formulated at pres-ent, five levels ofincreasing complexity are proposed: I, atomic:II. molecular: III, cellular; IV, tissue-system: and V. whole body.Although each level and its multiple compartments are distinct,biochemical and physiological connections exist such that themodel is consistent and functions as a whole. The model alsoprovides the opportunity to clearly define the concept ofa bodycomposition steady state in which quantitative associations existover a specified time interval between compartments at the sameor different levels. Finally, the five-level model provides a matrixfor creating explicit body-composition equations, reveals gapsin the study ofhuman body composition, and suggests importantnew research areas. .1,n J C/in Nutr 1992:56:19-28.

KEY WORDS Body composition, nutritional assessment.steady state

Introduction

The study of human body composition spans > 100 y andcontinues to be an active area of basic science and clinical re-search. Nearly every aspect of clinical nutrition, selected areaswithin many medical specialties, and components of exercisescience are touched on by the study of body composition.

Information related to body composition is accumulatingrapidly and is extending our knowledge ofhuman biology. Mostofthis information is now categorized as technical or biological.The technical category includes the many classic and continuallyemerging new body-composition methods. Although no sys-tematic classification for body-composition methodology hasbeen proposed. informal groupings are often published. such asdilution techniques and neutron-activation analysis. which arebased on a physical principle or other characteristics ofthe tech-niques involved. The biological category includes informationon the study ofhow growth, development, pregnancy. lactation,aging. exercise, and disease influence body composition.

Although the technical and biological categories would appearto encompass most body-composition information, a recentstudy ( 1 ) led us to appreciate a serious limitation of the field asit is now organized. We recognized that not all of the rapidlyaccumulating information emerging from body-compositionresearch could be satisfactorily included into the technical and

biological categories. For example, there are many mathematicalmodels that describe the relations between different componentsin healthy subjects [eg, total body water (TBW)/fat-free bodymass = 0.732] (2). This formulation indicates that some quan-titative associations exist that describe the relationships amongcompartments that are in equilibrium. Another example is pro-vided by the reconstruction of human chemical compartmentsand body weight (Bwt) from elements estimated in vivo by neu-tron-activation analysis ( 1 ). This suggests that relationships existnot only between individual components but between differentlevels of body composition as well.

Another problem is that investigators are frequently con-fronted with questions about terminology. For example: Arelipid-free body mass, fat-free body mass, and lean body mass(LBM) the same or different compartments? The lack of cleardefinitions for body-composition components has a subtle butserious consequence: many errors are evident in published body-composition equations and models because ofoverlap or omis-sion ofcomponents. In fact, we could find no clear approach todefining components and building multicompartment body-composition models in extensive reviews of previous literature.

Growing from these observations is the hypothesis that a thirdcentral category of body-composition research exists that untilnow has not been adequately formulated: the levels of bodycomposition and their organizational rules. This report presentsa comprehensive model ofhuman body composition consistingof five distinct levels of increasing complexity in which eachlevel has clearly defined components that comprise total Bwt.The five levels are I. atomic: II, molecular: III, cellular: IV. tissue-system: and V. whole body (Fig I).

The following section presents a detailed description of eachlevel and its associated components. In the next section the fea-tures or organizational rules of the model as a whole are de-scribed. Important concepts related to development of body-composition models and equations are presented in this portionofthe paper, and the widely appreciated but never formally de-fined concept ofa steady state ofbody composition is introduced.

I From the Obesity Research Center. St Lukes-Roosevelt Hospital,

Columbia University. College of Physicians and Surgeons. New York,NY.

2 Address reprint requests to Z-M Wang, Weight Control Unit. 4 I 1West I 14th Street, New York, NY 10025.

Received September 17, 1991.Accepted for publication December 12. 1991.

by guest on April 1, 2015ajcn.nutrition.org

Dow

nloaded from

Le1I(Atoalc)

20 WANG ET AL

FIG 1. The five levels of human body composition. ECF and ECS,extracellular fluid and solids, respectively.

TABLE 1Body composition on the atomic level (I) for the 70-kg ReferenceMan

Element Amount Percent ofbody weight

kg %

Oxygen 43 61Carbon 16 23Hydrogen 7 10Nitrogen 1.8 2.6Calcium 1.0 1.4Phosphorus 0.58 0.83Sulfer 0. 14 0.20Potassium 0. 14 0.20Sodium 0.1 0.14Chlorine 0.095 0.14Magnesium 0.019 0.027

Total 69.874 99.537

* Information based on reference 2 (modified).

Five-level model

Atomic (I)

The fundamental building blocks ofthe human body are atomsor elements. Ofthe 106 elements, 50 are found in the humanbody and their distributions in the various tissues and organsare well documented (2). Six elements (oxygen, carbon, hydro-gen, nitrogen, calcium, and phosphorus) account for > 98% ofBwt, and one element, oxygen, constitutes > 60% of total bodymass in the Reference Man (Table 1) (2). The remaining 44elements make up < 2% of Bwt.

The equation for Bwt, as defined in the atomic level of bodycomposition is

Bwt=O+C+H+N+Ca+P

+S+K+Na+Cl+Mg+R (1)

where R is the residual mass ofall elements present in amounts< 0.2% percent of Bwt (1).

Elemental analysis of humans is traditionally carried out incadavers or in biopsy specimens from selected tissues and organs.In addition, the whole-body content of most major elementscan now be measured directly in vivo: potassium by whole-bodycounting: sodium, chlorine, and calcium by delayed--y neutron-activation analysis (3); nitrogen by prompt-y neutron-activationanalysis (1 , 3): and carbon by inelastic neutron scattering (4).More than 98% of Bwt can now be reconstructed from elementsthat can be estimated in vivo, largely by neutron-activationtechniques. The atomic level is the foundation of body-corn-position analysis and is the starting point for the five levels wepropose.

P110/cell/ar (II)The 1 1 principal elements are incorporated into molecules

that form > 100 000 chemical compounds found in the humanbody. These molecules range in complexity and molecular weightfrom water to deoxyribonucleic acid. It is neither useful norpossible to measure all ofthese chemical compounds individuallyin living humans. The alternative used in body-composition re-search is to consider chemical compounds in categories of closely

related molecular species. The major components in present useare water, or aqueous (A): lipid (L); protein (Pro): mineral (M):and glycogen (G) (Table 2). Because some confusion exists inthese different categories. we now review the five chemical corn-ponents in detail.

Water. The most abundant chemical compound in the humanbody is water, which comprises 60% of Bwt in the ReferenceMan (2).

Protein. The term protein in body-composition research usu-ally includes almost all compounds containing nitrogen, rangingfrom simple amino acids to complex nucleoproteins. The mostwidely used representative stoichiometry for protein isCH 59N26O32S07 . with an average molecular weight of 2257.4and density of 1.34 g/cm3 at 37 #{176}C(1, 5).

G/icogen. The primary storage form of carbohydrate is gly-cogen. which is found in the cytoplasm of most cells. The prim-cipal distribution is in skeletal muscle and liver, which contain 1% and 2.2% of their respective wet weights in the form of

TABLE 2Body composition on the molecular level (II) for the 70-kg ReferenceMan*

Component Amount Percent of body weight

1cv %

WaterExtracellular 18 26Intracellular 24 34

LipidNonessential (fat) 12 17Essential 1.5 2.1

Protein 10.6 15Mineral 3.7 5.3

Total 69.8 99.4

* Glycogen. normally 400 g. is not included in the Reference Man.

Information based on reference 2.

by guest on April 1, 2015ajcn.nutrition.org

Dow

nloaded from

FIVE-LEVEL BODY COMPOSITION MODEL 21

glycogen ( I. 2). The stoichiometry of glycogen is (C6H10O5).with an average density of 1 .52 g/cm3 at 37 #{176}C( 1 , 2).

.tIiThl(ll. The term mineral describes a category of inorganiccompounds containing an abundance of metal elements (eg.calcium. sodium. and potassium) and nonmetal elements (eg.oxygen. phosphorus. and chlorine). Ash, a term similar to mm-eral. is the residue ofa biological sample heated for a prolongedperiod to > 500 #{176}C,and consists of the nonvolatile portion ofmineral compounds. Total body ash is slightly lower in weightthan mineral mass because of the loss of carbon dioxide fromsome carbonate groups and the release of tightly bound waterduring the heating period ( 1 . 2). Mineral is usually divided intotwo subcategories: osseous and extraosseous. Osseous mineral.the largest component of which is calcium hydroxyapatite([Ca3(PO4)2]3 Ca(OH)2). contains > 99% of total body calcium(TBCa) and 86% of total body phosphorus in the ReferenceMan (2). Other elements, such as potassium, sodium. and chlo-rime. are primarily found in extraosseous mineral.

Lipid. Among the five principal chemical components on themolecular level, lipid is the most confusing because the termslipid and fat are used interchangeably. even though strictlyspeaking they refer to different compartments. The traditionaldefinition oflipid refers to a group ofchemical compounds thatare insoluble in water and very soluble in organic solvents suchas diethyl ether, benzene, and chloroform (6, 7). About 50 dif-ferent lipids are recognized in humans, and these are divided byorganic chemists into five subcategories: 1) simple lipids (in-cluding triglycerides and waxes): 2) compound lipids (eg. phos-pholipids and sphingolipids): 3) steroids: 4) fatty acids: and 5)terpenes (6).

The simple lipid, triglyceride, contains three fatty acids ester-ifled to glycerol. The term fat is synonymous with triglycerideand therefore fat is clearly a subcategory of total lipid (6. 7). Acommon error is to confuse the terms fat and lipid, which canlead to errors in constructing models of body composition. Inthe adult. 90% oftotal body lipid is fat (2).

Lipids can also be classified physiologically into two groups:essential (Le) and nonessential (Ln) (2). Essential lipids. such assphingomyelin and phospholipids, serve important functionssuch as forming cell membranes. The nonessential lipids, largelyin the form of triglyceride, provide thermal insulation and astorage depot ofmobilizable fuel. About 10% oftotal body lipidis essential and 90% is nonessential in the Reference Man (2).

Although essential and nonessential lipids are structurally andphysiologically different. their solubilities in organic solvents aresimilar and it is difficult to clearly separate them even in vitro(6. 7). An approximate separation can be accomplished by carefulselection ofthe type oftissue analyzed, the extraction time andtemperature, and particularly the type of solvent used (6). Sol-vents such as petroleum or ethyl ether are usually used alone toextract nonessential lipids, mainly the neutral fat or triglyceride.The remaining lipids. which are primarily essential. can be ex-tracted by using binary or ternary solvent mixtures such as 45%chloroform. 10% methanol, and 45% heptane (8).

The fatty acid profile oftriglyceride varies with diet. anatomicsite. and other factors. but the generally accepted representativestoichiometry found in adult humans is C51H98O6, with an av-erage molecular weight of 806 and a density of 0.900 g/cm3 at37#{176}C(2). The stoichiometry oftotal lipid in humans could notbe found in a review ofprevious studies.

The equation for Bwt as defined by the molecular level ofboth composition is

Bwt= L+A+Pro+M+G+R (2)

where R represents residual chemical compounds not includedin the five main categories and that occur in quantities of < I %oftotal Bwt (1).

On the molecular level, three related equations can also bedefined as follows. Dry Bwt consists ofthe anhydrous chemicalcomponents (Fig 2, left), and equation 2 can therefore be re-written as

Bwt = A + dry Bwt (3)

Dry Bwt according to this equation is the sum of L + Pro + M+ G + R.

Bwt = L + lipid-free body mass (4)

In equation 4 (Fig 2, right). lipid-free body mass is the materialremaining after extraction of a whole-body homogenate withappropriate organic solvents and optimum conditions. Thuslipid-free body mass can be expressed as the combined weightof A + Pro + M + G + R.

As fat accounts almost entirely for total body nonessentiallipid, then

Bwt = fat + FFM = Ln + FFM (5)

where FFM is fat-free body mass, which represents the combinedweights of Le + A + Pro + M + G + R.

A similar term to fat-free body mass is LBM. The early def-inition of lean body mass suggested included at least five corn-ponents: water, protein. mineral. glycogen, and an unspecifiedamount of essential lipid (9. 10). More recently, most investi-gators have used the terms LBM and FFM interchangeably. al-though some debate still prevails about whether or not these arethe same or different compartments. Our suggestion is that LBMand FFM henceforth be considered synonymous on the basis ofthe following reasoning.

In equation 4 we clearly define two fractions ofBwt, lipid andlipid-free body mass. The lipid fraction consists oftwo portions.essential and nonessential or fat. Accordingly,

FIG 2. Body-composition model on the molecular level (II). FFM.fat-free body mass: LFM. lipid-free body mass: and Le and L. essentialand nonessential lipids. respectively.

by guest on April 1, 2015ajcn.nutrition.org

Dow

nloaded from

22 WANG ET AL

* Ln, nonessential lipid or fat; and Le, essential lipid.

Body weight Ln + LBM = fat + FFM

in which both LBM and FFM are the sum ofessential lipid pluslipid-free body mass and the remaining Bwt is nonessential lipidor fat. All ofthe terms ofthe molecular level are consistent witheach other when defined according to these guidelines and asshown in Table 3.

At present the main direct techniques available for estimatingcomponents on the molecular level are for water and mineral.TBW can be measured by several well-established isotope-di-lution techniques ( 10. 1 1), and osseous mineral can be estimatedby whole-body dual-photon absorptiometry (12). The remainingcomponents ofthe model must be estimated indirectly by usingmeasurements included in one of the other four levels. For ex-ample, protein can be determined from total body nitrogen atthe atomic level by making two assumptions: that all of bodynitrogen is in protein and that 16% of protein is nitrogen (1).Another example, total body fat can be calculated from bodydensity, which is a dimension at the whole-body level, by as-suming that fat and FFM have respective densities of0.900 and1.100 g/cm3 (9, 10).

The molecular level of body composition is the conceptualfoundation for the higher levels that follow. Also, the molecularlevel connects the study of body composition to other researchareas, notably biochemistry.

Cellular (III)Although the human body can be divided into different corn-

ponents at the molecular level, it is the assembly of these corn-ponents into cells that creates the living organism. The coordi-nated functions and interactions between cells are central to thestudy of human physiology in health and disease. The cellularlevel is therefore an important area of body-composition re-search.

The human body is composed of three main compartmentson the cellular level: cells, extracellular fluid, and extracellularsolids. Each ofthese compartments is now described in additionaldetail.

Cells. The cells possess the characteristics of life includingmetabolism, growth, and reproduction. Although the l018 cellsof the adult human body share many properties in common,there are great variations in size, shape, elemental and molecularcomposition, metabolism, and distribution. Cells are adapted tospecific functions, such as support, electrical conduction, andcontraction. Based on these differences, four categories of cellscan be defined: connective, epithelial, nervous, and muscular (13).

(6) Connective cells include three groups: loose, dense. and spe-cialized ( 1 3). Adipocytes. or fat cells, are a type of loose con-nective cell in which fat is stored. Bone cells. the osteoclasts andosteoblasts. and blood cells are representations of specializedconnective cells.

Muscle cells include striated skeletal, smooth. and cardiac.The striated skeletal muscle cells are the foundation of humanmovement and account for a large fraction ofbody weight. Cellsconsist offluid and solid components, the intracellular fluid andsolids.

E.vtracel/u/ar fluid. The nonmetabolizing fluid surroundingcells that provides a medium for gas exchange, transfer of nu-trients, and excretion of metabolic end products is referred toas the extracellular fluid.

Extracellular fluid, which is 94% water by volume, is dis-tributed into two main compartments: plasma in the intravas-cular space and interstitial fluid in the extravascular space.Plasma and interstitial fluid account for 5% and 20% of Bwtin the Reference Man (2), respectively.

Extrace/lular solids. Extracellular solids are also a nonmetab-olizing portion of the human body that consists of organic andinorganic chemical compounds. The organic extracellular solidsinclude three types of fiber: collagen, reticular, and elastic (13).Both collagen and reticular fibers are composed ofcollagen pro-tein whereas elastic fibers are formed from the protein elastin.

The inorganic extracellular solids represent 65% ofthe drybone matrix in the Reference Man (2). Calcium. phosphorus,and oxygen in bone are the main elements ofthe inorganic ex-tracellular solids that are incorporated into calcium hydroxy-apatite ( 1 ). Other inorganic components are also present in ex-tracellular solids. including bicarbonate, citrate, magnesium, andsodium (1. 2).

From the previous discussion, the cellular level of body corn-position can be accurately described by the equations

Bwt = CM + ECF + ECS

CM = muscle cells + connective cells

+ epithelial cells + nervous cells

ECF = plasma + 1SF

ECS = organic ECS + inorganic ECS

(7)

(8)

(9)

(10)

where CM is cell mass, ECF is extracellular fluid, ECS is extra-cellular solids. and 1SF is interstitial fluid. However. becausemost components in equations 7-10 cannot be measured in

TABLE 3Different body-composition terms on the molecular level (II)

Lipi ds*

Ln Le Water Protein Mineral Glycogen

Bodyweight x X X X X xDry body weight x X x x xLipid-free body mass x x x xFat-free body mass x x x x xLean body mass x x x x x

by guest on April 1, 2015ajcn.nutrition.org

Dow

nloaded from

Inorganic Cell Residual

Inorganic ECI ResIdual

Inorganic ECS

Total

BodyMineral

FIG 4. Relationship between total body mineral and inorganic solids.ECF. extracellular fluid: ECS, extracellular solids.

FIG 3. Relationship between body fluids. ECF. extracellular fluid:ECW. extracellular water: ICF. intracellular fluid: ICW. intracellular wa-ter: R, and R, . extracellular and intracellular residual: and TBW. totalbody water.

FIVE-LEVEL BODY COMPOSITION MODEL 23

vivo at present. the following equation is suggested as a practicalalternative for Bwt at the cellular level

Bwt = fat cells + BCM + ECF + ECS ( Ii)

where BCM is body cell mass. 13CM is a portion of total cellmass that according to Moore et al ( 1 1 ) is the working. energy-metabolizing portion ofthe human body in relation to its sup-porting structure. Hence, BCM includes the protoplasm in fatcells hut does not include the stored fat, which occupies 85% to90 offat cell weight. Although no present method can directlymeasure BCM. it is a widely used term and is assumed to berepresented by exchangeable or total body potassium (TBK) ( I 1).A deviation must be noted in equation 1 1 in that BCM and fatcells share in common the nonfat portion of adipoctes andtherefore overlap by 1 kg in the Reference Man (2).

The fluid compartments at this level can also be related toTBW as shown in Figure 3. According to this model. ECW andICW are extracellular and intracellular water. and Re and Riare nonaqueous residual extracellular and intracellular solids.

Another relation at the cellular level is between total bodymineral and inorganic solids (Fig 4). Each ofthe three compo-nents in equation 7 contribute to total mineral, inorganic celland extracellular fluid residual, and the inorganic portion ofextracellular solids.

Of the three primary compartments at the cellular level. thevolume of extracellular fluid and its plasma subcompartmentcan be quantified directly by dilution methods ( 10). In contrast.no direct methods are yet available for estimating either cellmass or extracellular solids. Indirect methods ofevaluating somecompartments are available, such as extracellular solids estimatedfrom TBCa measured by neutron-activation analysis (ECS= TBCa/0. 1 77) ( 10). Another example is the calculation of BCMfrom TBK [BCM (in kg) = 0.00833 X TBK (in mmol)] ( 1 1)

Because the cellular level is the first level at which character-istics ofthe living organism appear. it occupies a central positionin connecting the inanimate features of body composition atthe lower levels with those of the animate features of tissues.organs. and intact humans at the higher levels. Despite its im-portance in the study of human body composition. very littleresearch has been directed at this level, perhaps because of thedifficulty in quantifying some ofthe compartments.

Tissue-System (I J )At the cellular level the human body is composed of cells.

extracellular fluid. and extracellular solids. These three corn-ponents are further organized into tissues. organs. and systems-the fourth level of body composition.

iissl1Ls. Generally. tissues contain cells that are similar inappearance. function. and embryonic origin. All ofthe diversetissues ofthe body can be grouped into four categories: muscular.connective. epithelial. and nervous ( 13).

Bwt at the tissue level ofhody composition is defined as

Bwt = muscular tissue + connective tissue

+ epithelial tissue + nervous tissue (/2)

Three specific tissues are particularly important in body-corn-position research: hone. adipose. and muscular, which togethercomprise 75 ofBwt in the Reference Man (2).

Bone is a specialized form of connective tissue that consistsof bone cells surrounded by a matrix of fibers and ground sub-stance. The distinguishing feature of bone is that the groundsubstance is calcified and accounts for 65 ofdry bone weight(2). The calcified ground substance is mainly hydroxyapatite([Ca3(P04)2]3Ca(OH),) and a small amount of calcium car-bonate (14).

Adipose tissue is another type of connective tissue made upof fat cells (adipocvtes) with collagenous and elastic fibers. fi-broblasts. and capillaries. Adipose tissue can be divided intofour types according to its distribution: subcutaneous. visceral(ie. loosely surrounds organs and viscera). interstitial (ie. inti-mately interspersed among the cells oforgans). and yellow mar-row (2). Muscle tissue can be subdivided into striated skeletal,smooth. and cardiac tissues (2).

Oq.aiis. The organs consist oftwo or more tissues combinedto form large functional units such as skin, kidney, and bloodvessels.

Siste,ns. Several organs whose functions are interrelated con-stitute an organ system. For example. the digestive system iscomposed of many organs. including the esophagus. stomach,intestine. liver. and pancreas. Each organ. such as the stomach.

by guest on April 1, 2015ajcn.nutrition.org

Dow

nloaded from

24 WANG ET AL

contains several kinds oftissue (muscular. connective. epithelial.and nervous) and each tissue is made up of many cells andextracellular material.

There are nine main systems in the human body. hence Bwtat the system level of body composition can be defined as

Bwt = musculoskeletal + skin + nervous

+ circulatory + respiratory + digestive + urinary

+ endocrine + reproductive systems (13)

Although Bwt can be expressed accurately on the tissue-systemlevel, most components in equations 12 and 13 cannot be mea-sured in vivo at present. The following equation is suggested asa practical alternative

Bwt = adipose tissue + skeletal muscle

+ bone + viscera + blood + R (14)

where the five components account for 85% and R accounts forthe remaining 1 5% of Bwt in the Reference Man (Table 4) (2).

The tissue-system level is complex and interfaces with severalbranches ofhuman biology, including histology and histochem-istry at the tissue level and anatomy and physiology at the organand system level. Physicians, nutritionists, and exercise physi-ologists focus much oftheir interest in body composition at thetissue-system level.

Although a great deal of information is available at this level,most of it comes from cadaver studies or tissue biopsies. Thereare only a few in vivo direct methods that can be used to estimatethe major compartments at the tissue-system level. An exampleis computerized axial tomography, which can directly determinethe volume of subcutaneous and visceral adipose tissue (15).Some indirect techniques are also available at this level. such asestimation ofskeletal muscle mass from 24-h urinary creatinineexcretion or from TBK and nitrogen content by neutron-acti-vation analysis (16, 17).

i/i Ito/c body (I)Both humans and some primates have similar body compo-

sitions at the atomic, molecular, cellular, and tissue-system levels.It is at the whole-body level, however, with its complex char-acteristics that distinguishes humans from all other primates. Inaddition, many biological, genetic, and pathological processeshave an impact not only at the first four levels but also on thehuman body as a whole.

The whole-body level of body composition concerns bodysize, shape, and exterior and physical characteristics. There are 10 suggested dimensions at the whole-body level (18).

I) Stature: This is a major indicator ofgeneral body size andskeletal length.

2) Segment lengths: Many segment lengths are used in thestudy ofbody composition, the most common ofwhich are lowerextremity length, thigh length, calflength, shoulder-elbow length,and elbow-wrist length.

3) Body breadths: Body breadths are a measure ofbody shape,skeletal mass, and frame size. The sites most widely used arethe wrist, elbow, ankle, knee, and biiliac.

4 ) Circumferences: The circumferences are useful indicatorsof body density, FFM, adipose tissue mass, total body proteinmass. and energy stores. The most widely used circumferencesare upper arm, waist (abdominal), and thigh.

TABLE 4Body composition on the tissue-system leveand organs ofthe 70-kg Reference Man*

I (IV) for principal tissues

Tissue or organ Amount Percent of body weight

&v %

Skeletal muscle 28Adipose tissue

Subcutaneous 7.5Visceral 5Interstitial 1Yellow marrow I .5

Bone 5Blood 5.5Skin 2.6Liver 1.8Central nervous system I .4Gastrointestinaltract 1.2Lung I

40

1 17.11.42.17.17.93.72.621.71.4

* Information based on reference 2 (modified).

5) Skinfold thicknesses: Skinfolds represent a double layer ofadipose tissue and skin at specific anatomic locations. Triceps.subscapular, calf (medial), and abdominal are the most corn-monly used sites. Skinfold thickness provides a simple methodof estimating fatness and the distribution of subcutaneous adi-pose tissue. Numerous equations for the prediction of body fathave been developed that make use of skinfold thicknesses.

6) Body surface area (BSA): The total BSA is an exteriorcharacteristic that is often used to estimate basal metabolic rateand FFM.

7) Body volume: The total body volume is an important in-dicator of body size and is used to calculate body density.

8) Bwt: One ofthe simplest and most important morphologicindicators. Bwt is used in screening for growth rate, obesity. andundernutrition. The Bwt equation that defines the whole-bodylevel is

Bwt = head weight + neck weight + trunk weight

+ lower extremity weights + upper extremity weights (15)

9) Body mass index: Bwt and stature can be combined toform indices that correlate with total body fat. The best knownof the indices is body mass index (body weight/stature2. in kg!m2), which is often used in obesity studies as a measure of fatness( 19). However, more complex and population-specific indices,such as the Fels index (Bwt2/stature33), often correlate betterwith total body fat ( 18).

10) Body density: The density of the human body, derivedfrom Bwt and volume, is widely used to indirectly estimate totalbody fat and FFM (9, 10) and is defined at the molecular levelas

1/Db = fFat/DF + fFFM/DM (16)where Db. DF, and DFFM are the densities (in g/cm3) ofthe totalbody, fat, and fat-free body, respectively, and f represents thefractions of Bwt as fat and FFM, respectively (20). Similar equa-tions for total body density based on individual components atthe cellular, tissue-system, and whole-body levels can also bewritten.

by guest on April 1, 2015ajcn.nutrition.org

Dow

nloaded from

FIVE-LEVEL BODY COMPOSITION MODEL 25

TABLE 5Some related but distinct components on different levels

Atomic level Molecular level Cellular level

Tissue-syste m level

Tissue level Organ level

Total hod calciumand phosphorus

Mineral Extracellular solids Bone Skeleton

Total body carbon Lipid and fat Fat cells Adipose tissue

Skeletal muscle cells Skeletal muscle tissue Skeletal muscle

It is clear that any major changes in body composition on theother four levels will manifest themselves on the whole-bodylevel. Conversely. most differences at the whole-body level arerelated to changes in composition on the other four levels. Thislatter relation is the foundation for estimating the componentsof the other four levels by using measurements at the whole-body level. Most indicators at the whole-body level are simplerand easier to perform than are measurements at the other fourlevels. thus the techniques at this level are often well suited forlarge-scale studies or for field work.

Features of the model

The five-level model provides a structural framework forstudying human body composition that goes beyond an mdi-vidual compartment or level. In this section we describe someof the features of the five-level model as a whole.

Distinctions and connections bettieen diflirent levelsAn essential aspect of the model is that the levels themselves

are distinct and have unique properties that should not be con-fused with one another.

I) On the atomic level, there are no special elements or anyfundamental differences between the human body and the in-organic world. although the ratio ofelements to each other varies.

2) On the molecular level, the human body is differentiatedfrom the inorganic world because ofthe appearance of complexorganic compounds such as lipid and protein.

3) On the cellular level, the human body is distinct from thenonliving world because ofthe appearance ofcells that have thecharacteristics of living organisms.

4) On the tissue-system level, the human body is differentfrom the lower animal world because ofthe appearance of tissues,organs. and systems having complex structures and functions.

5) On the whole-body level, the human body is differentiatedfrom all other primates because ofthe presence ofdistinct mor-phological features.

Although these distinct properties exist for each of the fivelevels, linkages are also present that are clearly recognizable inthe context ofthe five-level model. An example is that cells thatappear first on the cellular level have many ofthe characteristicsofliving organisms such as membrane transport, energy metab-olism. and enzymatic processes. These characteristics ofthe cellare still maintained at the tissue-system and whole-body levels.Each higher level is thus unique but maintains some ofthe char-acteristics of the level below it.

Recognition ofdistinct levels and their connections can revealgaps in present body-composition information and suggest a

direction for future research efforts. For example. it is knownthat many biological factors including growth, development, se-nescence. race, sex, nutritional status, exercise level, and thepresence of disease all have important effects on body compo-sition. However, most studies ofbody composition in these areasare limited in scope, focusing on only a few components at oneor two levels and thereby failing to appreciate the connectionsbetween levels. For example, most previous obesity studies werelimited to anthropometric changes (at the whole-body level) andalterations in fat mass (at the molecular level). Very few studieshave investigated how obesity influences the other levels of bodycomposition or more importantly the coordinated changes thatoccur throughout all five levels with increasing Bwt.

Distinctions and connections between different componentsAn important feature ofthe model is that every major corn-

ponent has a clear definition and can be included in one of thefive levels. Each ofthese components has unique properties andyet maintains relationships with other components at the sameand different levels.

It was not unusual in earlier studies for related componentsto be confused with each other, particularly ifthey were on dif-ferent levels. An example of three sets of commonly confusedcomponents is presented in Table 5. In the first set, TBCa andphosphorus. mineral, extracellular solids, bone tissue, and skel-eton, are related compartments but belong to different levelsand have distinct differences from each other:

1) Calcium and phosphorus, and mineral: Most ofTBCa andphosphorus exist in mineral although there is some phosphorusin protein and lipid (eg, DNA, RNA, and phospholipid). On theother hand, in addition to calcium and phosphorus, mineralcontains other elements (eg, carbon, oxygen, hydrogen, mag-nesium, and sodium).

2) Mineral and extracellular solids: Most oftotal body mineralis in extracellular solids although there still is a small amountof mineral in cells and extracellular fluid. On the other hand,in addition to the mineral in the form of inorganic material.extracellular solids contain organic solids such as collagen, re-ticular fibers, and elastic fibers.

3) Extracellular solids and bone tissue: Most of total bodyextracellular solids are in the form ofbone tissue although therestill is a small amount ofextracellular solids in other tissues (eg,in skeletal muscle). On the other hand, in addition to extracellularsolids. bone tissue contains bone cells and extracellular fluid.

4) Bone tissue and skeleton: Bone tissue constitutes the ma-jority of the skeleton although the latter also includes skeletalcartilage, periarticular tissue adhering tojoints, and red and yel-low marrow.

by guest on April 1, 2015ajcn.nutrition.org

Dow

nloaded from

26 WANG ET AL

Another example of related but distinct components is totalbody carbon (level I). lipid and fat (level II), fat cells (level III),and adipose tissue (level IV) (Table 5). These terms are oftenconfused with each other, a problem that the five-level modelhelps to resolve.

The third and final example in the table is the distinctly dif-ferent but related components skeletal muscle cells, skeletalmuscle tissue, and intact whole skeletal muscles. The modelthus demonstrates that differences and relations exist betweencomponents on each of the five levels. It is therefore advisableto develop equations for body weight, volume, or density thatinclude components from the same level in order to avoid overlapor omission of some components.

Steady-state OfbOdI compositionThe concept of a steady state is important not only in bio-

chemistry, physiology, and other classic scientific disciplines butalso in body-composition research. The meaning of a steadystate of body composition can be defined in the context of thefive-level model: A steady-state or dynamic homeostasis existsduring a specified time period if Bwt and the mass of variouscomponents on different levels is maintained relatively constant.

The important implication of a steady state is that there arestable proportions among the different components on the samelevel. For example, on the molecular level the average ratio oftotal body water content to FFM is relatively constant in healthysubjects (ie, total body water/FFM = 0.732) (2). On the atomiclevel the correlation between TBK and TBCa is reproduciblefor males [ie, TBK(g) 0. 1383 X TBCa(g) - 17. 1 ] (2 1). On thewhole-body level the relation between BSA (BSA, in m2) andBwt (kg) and stature (in rn) is also relatively constant such thatBSA = 0.007 184 x stature#{176}725 X body weight#{176}425(22).

There are also relatively constant proportions among the rel-evant components on different levels when body compositionis in a steady state. For example, total body protein/total body

nitrogen = 6.25 ( 1 ): BCM(kg)/TBK(mmol) = 0.00833 ( 1 1 ), andfat(kg) = [(4.95/Db) - 4.50] x Bwt (20).

The steady state of body composition indicates that althoughthere are so many components in the human body, and all ofthese components differ from each other, they are well organizedaccording to definable quantitative relations.

Quantitative hodi COlfl/)OsitiOfl relations

A primary aim of body-composition research is to estimatethe size ofeach compartment. although there are numerous in-dividual compartments ofclinical relevance that have not beenmeasured directly. An alternative is to estimate the unknowncomponents by establishing relationships to measurable corn-ponents. Body composition is relatively stable in healthy adults.and it is this property that enables investigators to establish thesereproducible relations or rules. The five-level model of bodycomposition affords a logical matrix within which to establishthe quantitative steady-state relations between known measur-able components and presently unmeasurable compartments.

Present research in developing body-composition equationsprimarily involves estimating one unknown component from ameasurable component. The five-level model suggests the pos-sibility of reconstructing Bwt and volume by writing simulta-neous equations that exploit steady-state relations between sev-eral measurable and unknown components. An example is thecalculation of the five major chemical components and Bwt atthe molecular level from six elements (carbon, nitrogen, sodium.potassium. chlorine, and calcium) measured by in vivo neutron-activation analysis ( 1). Until recently the concept of recon-structing whole levels ofbody composition from multiple com-ponents was limited and the studies were fragmentary. The five-level model defines explicitly the equations for Bwt at each leveland presents the challenge of developing more complex andcomprehensive body-composition equations.

TABLE 6The relation between direct and indirect body-composition measurements organized by the five-level model

Direct

Indirect

Atomic level Molecular level Cellular level Tissue-system level Whole-body level

Atomic level TBP = (0.456 X TBCa) Pro = 6.25 X TBN BCM = 0.00833 X TBK SM = 0.0196 X TBK Bwt 0 + C + H + NTBO. TBC. TBH, TBN. + (0.555 x TBK) FFM = TBK/68.l ECS = TBCa/0.l77 - 0.0261 X TBN + Ca + P + K + Na

TBK. TBCa. TBNa. ECF = (0.9 X TBCI)/ + Cl + RTBP. TBCI. Nae. Ke Plasma Cl

Molecular level FFM = TBW/O.732 SM = 1 1 .8 X Cr Bwt = L + A + ProTBW, mineral. creatinine. FFM = 24.1 X Cr + 10.1 + M + G + R

3-MH + 20.7Cellular level Bwt = CM + ECF

ECF. plasma volume + ECSTissue-system level Bwt = adipose tissue

Volume of subcutaneous + skeletal muscle

and visceral adipose + bone + vi5Cerstissue + blood + R

Whole-body level TBK = (27.3 X Bwt) Fat% = (4.95 X BV/ ECF = 0.135 X Bwt SM = 5(0.0553 Body surface = 0.007184Bwt, S. By, circumference. + (11.5 X 5) - (21.9 Bwt - 4.5) X 100 + 7.35 X CTG2 + 0.0987 X 50725 x BW#{176}425

skinfold x Age) + 77.8 x FG2 + 0.0331x CCG2) - 2445

* A. water (kg): BCM. body cell mass (kg): BV. body volume (L): Bwt. body weight (kg): CCG. corrected medial calf girth (cm): CM. cell mass (kg): Cr, 24-h urinecreatinine (g): CTG. corrected thigh girth (cm): ECF. extracellular fluid (kg): ECS. extracellular solids (kg): FFM. fat-free body mass (kg): FG. forearm girth (cm): G.glycogen(kg): Ke. exchangeable potassium: L. lipid(kg): M. Mineral(kg): 3-MH. 24-h urine 3-methylhistidine: Nae. exchangeable sodium: plasma Cl. plasma concentrationofchlorine (mmol/L): Pro. protein (kg): R. residual (kg): S. stature (cm): SM. skeletal muscle (kg): TB. total body element (kg): and TBW. total body water (kg).

by guest on April 1, 2015ajcn.nutrition.org

Dow

nloaded from

FIVE-LEVEL BODY COMPOSITION MODEL 27

Re/atiomi to 1flCt/lOdOl()tl

At present. body-composition methods are primarily cate-gorized into technique-specific groupings such as dilution meth-ods and neutron-activation analysis. According to the five-levelmodel, however, the methods can be organized in a more sys-tematic fashion.

Direct measurement !fitt/lOds. There are some direct methods,such as anthropometric, biochemical, and radioisotopic tech-niques that can be used to estimate components of body com-position. Direct methods can be organized according to the five-level model as follows:

1) On the atomic level, TBK can be directly determined bywhole-body #{176}Kcounting ( 12): total body sodium. chlorine.phosphorus, and calcium by delayed-y neutron activation (3):total body nitrogen by prompt--) neutron activation (23): andtotal body carbon by inelastic neutron scattering (4).

2) On the molecular level, TBW can be directly estimated byseveral isotope-dilution techniques (10), and osseous mineralcan be quantified by dual-photon absorptiometry (24).

3) On the cellular level, extracellular fluid (and plasma vol-ume) can be directly determined by several isotope-dilutiontechniques (25).

4 ) On the tissue-system level, the volumes of subcutaneousand visceral adipose tissue can be directly determined by com-puterized axial tomography and by magnetic resonance imagingtechniques (26).

5) On the whole-body level, anthropometric indices such asBwt, body volume, stature, circumferences, and skinfold thick-nesses can be estimated directly (18).

According to this analysis there are not many direct methodsused in the study ofhurnan body composition. Moreover, mostofthe direct methods are concentrated on the atomic and whole-body levels. There are only a few direct techniques on the mo-lecular, cellular, and tissue-system levels.

Indirect neasuremnent methods. These estimate unknowncomponents of body composition by combining direct mea-surernent techniques with the established steady-state relation-ship between the directly measurable and unknown components.Indirect methods greatly expand the number of body compart-rnents that can be evaluated. At present, some important corn-partments can be assessed only by indirect methods. For ex-ample, although total body fat is a major compartment of in-terest. there are no practical methods ofdirectly evaluating thefat compartment in vivo. All ofthe presently used methods areindirect and based on direct measurements at different levels asfollows:

1) from direct method on the atomic level (10), fat = Bwt- TBK(mmol)/68. 1 : 2) from direct method on the molecularlevel (2, 10), fat = Bwt - TBW/0.732: 3) from direct methodon the molecular and whole-body levels, fat = 2.057 X BV- 0.786 x TBW - 1 .286 X Bwt, where BV is body volume inliters (20); and 4) from direct methods on the whole-body level,fat=4.95XBV-4.5OXBwt,andfat=0.7l5XBwt- 12.1x stature2 (in m) (19, 20).Thus it can be seen that indirect methods are not only based onthe direct methods, but also are dependent on the steady-stateproportions between known and unknown components as de-termined in sample populations.

Direct and indirect body-composition methodology can beoutlined according to the five-level model as shown by the ex-

The Study of Body Composition

Components on Levels I, II, III, IV, and VBody composition rules

I _ IL Methodology j IBiological effects

FIG 5. The three areas ofbody-composition research.

amples presented in Table 6. The table demonstrates that mostof the principal elements and anthropometric indices can bedirectly measured and that many of the indirect methods havebeen developed from the direct methods on the atomic andwhole-body levels. respectively ( I 0, 17, 22, 25, 27-29). Con-versely, the table shows that there are only a few direct methodson the cellular and tissue-system levels, so the relevant indirectmethods are also very limited. This is one ofthe weak areas inbody-composition methodology and could constitute an im-portant topic for future research.

Definition of body composition research

The study of body composition spans > 100 y, and the termbody composition is widely used. However, it is unclear whatthis branch of science represents and what exactly is meant bythe term body composition. The five-level model presented inthis paper not only builds an appropriate structure for body-composition research. but is conducive to clearly define humanbody composition as a branch of human biology that studiesvarious body compartments and their quantitative steady-staterelations or rules. Body-composition research includes three in-terconnecting areas: studying the proportions of various corn-ponents and their steady-state associations among the atomic,molecular. cellular, tissue-system, and whole-body levels: study-ing the methods of measuring various components in vivo; andstudying the influences of biological factors on various levelsand components (Fig 5).

Conclusion

The five-level model grows from a need to organize both therapidly developing methodologies and physiological conceptsthat relate to the study ofhuman body composition. The modelis intended to be a foundation on which future studies can refineor expand selected definitions or equations. The five-level modelserves in this organizational capacity and also stimulates abroader view of body-composition research as a whole. C]

References

I . Heymsfield SB, Waki M, Kehayias J. et al. Chemical and elementalanalysis ofhumans in vivo using improved body composition models.Am J Physiol 199l:26l:E190-8.

2. Snyder WS. Cook Mi. Nasset ES, Karhausen LR. Howells OP.Tipton lH. Report of the task group on Reference Man. Oxford:Pergamon Press, 1984.

by guest on April 1, 2015ajcn.nutrition.org

Dow

nloaded from

28 WANG ET AL

3. Cohn SH, Dombrowski CS. Measurement of total-body calcium.sodium, chlorine. nitrogen. and phosphorus in man by in-vivo neu-tron activation. I NucI Med 197 1:12:499-505.

4. Kehayias Ii. Heymsfield SB. Dilmanian FA. Wang I, Gunther DM,Pierson RN. Measurement ofbody fat by neutron inelastic scattering:comments on installation. operation. and error analysis. In:Yasumura 5, Harrison JE, McNeill KG. Woodhead AD. DilmanianFA. eds. In vivo body composition studies. New York: Plenum Press,1990:317-25.

5. Kleiber M. The fire oflife. Huntington. NY: Robert E Kreiger. 1975.6. Gurr MI. Harwood IL. Lipid biochemistry. 4th ed. London: Chap-

man and Hall, 1991.7. Diem K. Constituents of living matter. 6th ed. Ardsley. NY: Geigy

Pharmaceuticals, 1962.8. Knight OS. Beddoe AH. Streat SI. Hill GL. Body composition of

two human cadavers by neutron activation and chemical analysis.Am I Physiol 1986:250:El79-85.

9. Behnke AR, Wilmore JH. Evaluation and regulation ofbody buildand composition. Englewoocl Cliffs. NJ: Prentice-Hall. 1974.

10. Forbes GB. Human body composition. New York: Springer-Verlag.1987.

1 1. Moore FD. Olesen KH. McMurray ID. Parker HV, Ball MR. BoydenCM. The body cell mass and its supporting environment. Philadel-phia: WB Saunders. 1963.

12. Heymsfield SB. Wang J. Aulet M, et al. Dual photon absorptiometry:validation of mineral and fat measurements. In: Yasumura 5,Harrison JE, McNeill KG. Woodhead AD. Dilmanian FA, eds. Invivo body composition studies. New York: Plenum Press, 1990:327-37.

13. Jacob SW, Francone CA. Lossow WI. Structure and function inman. 4th ed. Philadelphia: WB Saunders. 1978.

14. Armstrong WD, Singer L. Composition and constitution of themineral phase ofbone. Clin Orthop 1965:38:179-90.

15. Kvist H. Chowdhury B. Grangard U. Tylen U. Sjostrom L. Totaland visceral adipose-tissue volumes derived from measurements withcomputed tomography in adult men and women: predictive equa-tions. Am I Clin Nutr 1988:48:1351-61.

16. Heymsfield SB. Arteaga C. McManus C. Smith J, Moffitt S. Mea-surement ofmuscle mass in humans: validity ofthe 24-hour urinarycreatinine method. Am J Clin Nutr 1983:37:478-94.

17. Burkinshaw L. Hill GL. Morgan DB. Assessment ofthe distributionofprotein in the human body by in-vivo neutron activation analysis.International Symposium on Nuclear Activation Techniques in theLife Sciences. Vienna: International Atomic Energy Association.1978.

18. Lohman T. Roche AF. Martorell R. eds. Anthropometric standard-ization reference manual. Champaign, IL: Human Kinetics Books.1988.

I 9. Garrow IS, Webster I. Quetelets index (W/H2) as a measure offatness. Int I Obes 1985:9:147-53.

20. Sin WE. Body composition from fluid spaces and density: analysisof methods. In: Brozek I, Henschel A, eds. Techniques for measuringbody composition. Washington: National Academy ofScience. 1969.

2 1. Ellis KI. Cohn SH. Correlation between skeletal calcium mass andmuscle mass in man. I Appl Physiol 1975:38:455-60.

22. DuBois D. DuBois EF. Clinical calorimeter. A formula to estimatethe approximate surface if height and weight be known. Arch IntMed 1916:17:863-71.

23. Cohn SH, Vartsky D. Yasumura S. Vaswan AN, Ellis KJ. Indexesofbody cell mass: nitrogen versus potassium. Am J Physiol 1983:244:E305-l0.

24. Pierson RN Jr. Wang I. Heymsfield SB, Dilmanian FA. Weber DA.High precision in-vivo neutron activation analysis: a new era forcompartmental analysis in body composition. In: Yasumura S.Harrison JE. McNeil KG, Woodhead AD, Dilmanian FA, eds. Invivo body composition studies. New York: Plenum Press, 1990:3 17-25.

25. Yasumura 5, Cohn SH, Ellis KJ. Measurement ofextracellular spaceby total body neutron activation. Am J Physiol 1983:244:36-40.

26. Seidell IC. Bakker CJG. van der Kooy K. Imaging techniques formeasuring adipose-tissue distribution-a comparison betweencomputed tomography and 1.5-T magnetic resonance. Am I ClinNutr 1990:51:953-7.

27. Sin WE. The gross composition ofthe body. Advances in biologicaland medical physics. 1956:49:239-80.

28. Martin AD. Spenst LF. Drinkwater DT. Clarys IP. Anthropometricestimation of muscle mass in men. Med Sci Sports Exerc I 990:22:729-33.

29. Novak LP. Aging. total body potassium, fat-free mass, and cell massin males and females between the ages 18 and 85 years. J Gerontol1972:27:438-43.

by guest on April 1, 2015ajcn.nutrition.org

Dow

nloaded from