AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 63:323-329 (1984) Estimation of Body Fat in Female Rhesus Monkeys MARGARET L. WALKER, SUSAN M. SCHWARTZ, MARK E. WILSON, AND PAUL I. MUSEY Yerkes Regional Primate Research Center (M.L.W ,S.M.S., M.E. W.) and Department of Medicine (S.M.S., PI.M.%Emory University, Atlanta. Gebrgia 30322 KEY WORDS Body fat, Rhesus Monkey, Tritiated water, Morphometric measurement ABSTRACT Measurement of height (crown-rump length), body weight, and abdominal subcutaneous fat depth, based on skinfold thickness taken from 13 female rhesus monkeys comprising two age groups were correlated with body fat values derived from tritiated water determinations of total body water. The manner with which each measure was related to percent body fat differed as a function of age of the animal. In the young, nulliparous females, crown-rump length was the single best predictor of body fat, whereas in the older, multiparous females, skinfold thickness correlated most highly with body fat. When all measurements, including the Quetelet index [(Wtkt? x 1,0001, were combined statistically and regressed against percent body fat, a significant increase in predictive ability was obtained. When each age group was considered separately, the resulting equations again reflected the agegroup biases. In addition, as an internal check on the validity of the regression equations, an additional regression analysis was performed using morphometric data from selected animals in each age group. These equations yielded accurate estimates of body fat when compared to determinations made from total body water. These analyses indicate that the predictive accuracy of morphometric data is greatly enhanced by using these measurements in concert. Furthermore, the utility of such predictions is influenced by the specific physical characteristics of the subject population. The estimation of body fat in primates has generally been inferred from measurements of body weight (Durnin and Rahaman, 1967; Wilmore and Behnke, 1970;Young and Blondin, 1962). However, body weight does not necessarily provide an accurate estimate of body fat, since such factors as height, age, sex, muscle mass, and bone density can differentially contribute to the weight measurement and thereby reduce the accuracy of body fat estimates (Broxek, 1961). To overcome these obstacles, some investigators have employed more direct methods of assessing relative adiposity. Measurement of subcutaneous fat depth in macaques using skinfold calipers (Small, 1981; Walike et al., 1977) has been found to provide reliable data, although this technique is limited by its inability to measure fat depth in a range of less than 2 mm (Hansen, personal communication), and 0 1984 ALAN R.LISS. INC by changes in skinfold compressibility which occur with old age (Broxek and Kinzey, 1960). Ultrasonic measurement of fat has been employed successfully in humans (Bullen et al., 1965; Haymes et al., 1976), although it has yet to be validated for use in nonhuman primates. Other techniques, such as soft-tissue roentgenograms, carcass analysis, and water submersion, are unsuitable or economically impractical in nonhuman primates. One additional method, the measurement of total body water (Schloerb et al., 19501, has been adapted for use in macaques (Kamis and Latif, 1981; Walike et al., 19771, and this technique appears to provide valid estimates of body fat in monkeys. Received September 16, 1983; revised November 17, 1983; accepted November 28,1983. 324 M.L. WALKER, S.M. SCHWARTZ, M.E. WILSON, ANDP.1. MUSEY The ability to obtain accurate estimates of body fat is critically important to such areas of research as nutrition, development, and reproduction. For this reason, estimates of adiposity derived from morphometric data must be validated against actual determinations of body fat, since such measures alone can only provide data regarding relative differences among subjects. Furthermore, since body composition may change as a function of development, certain age-related factors must be taken into account. To this end, measurements of height (crown-rump length), weight, and abdominal subcutaneous fat depth based on skinfold thickness and ultrasound were made in 13 female rhesus monkeys comprising two age groups. The ability of these measures to estimate body fat accurately was established by means of multiple regression against percent body fat derived from total body water determinations. The regression equations thus generated provide a useful and practical index for the noninvasive assessment of percent body fat. MATERIALS AND METHODS Subjects Thirteen female rhesus monkeys (Macaca mulatta) housed at the Yerkes Primate Center Field Station served as subjects. These animals were selected because they represented two distinct groups with regard to size, age, and reproductive status. The two groups were (1)nulliparous females (n = 6) - mean age of 3.22 years ( & .02 years) and (2) multiparous females (n = 7) - mean age of 8.82 years ( & .52 years). All females were members of mixed-sex social groups housed in outdoor compounds. However, for the purposes of the present study, the animals were moved indoors 24 hours prior to the study onset and housed in individual cages. Upon completion of the experiment, the animals were returned to the original social groups. Since this study was conducted during the nonbreeding summer months, all subjects were anovulatory and none of the adult females was pregnant or lactating. Procedure The height, weight, and abdominal subcutaneous fat depth measurements were made on anesthetized animals (ketamine HC1, 5 mgkg) in order to minimize measurement error. All such measurements were made in duplicate. Except when otherwise stated, food and water were made available to the animals on a n ad lib schedule. Crown-rump Length. Height measurements (cm) were made using Vernier anthropometric calipers (Gneupel, Switzerland) according to the method outlined by Schultz (1929). One end of the calipers was placed on the cranial vertex (the highest point of the head in the midsagittal plane) and the other was positioned on the buttocks at the midline of the horizontal plane which bisects the ischial callosities. Subcutaneous Fat Depth. Subcutaneous fat depth (mm) was measured using large skinfold calipers (Cambridge Scientific Instruments, Cambridge, MD). The calipers were placed a t a point on the lower abdomen a t the midline, approximately 2 cm below the umbilicus. Ultrasound measurement (Scanoprobe, Model 731A, Ithaco, Ithaca, NY) of subcutaneous fat depth (mm) was made at the identical site on the abdomen. This site was chosen because preliminary data had revealed that triceps and subscapular sites yielded unreliable measurements due to the lack of fat deposits in these areas. Body Weight. Body weight measurements (Fairbanks Digital Scale, Model H90-7601, Colt Industries, St. Johnsbury, VT) were taken from animals that had been fasted for approximately 16 hours. These measurements were recorded in kilograms (kg). The height and weight measurements were also incorporated into,the Quetelet index [(body wt/htP x 1,000], since this index has been found to provide a n unbiased estimate of adiposity (Billewicz et al., 1962; Keys et al., 1972). Total Body Water Determination. This determination was conducted according to a modification of procedures described earlier (Kamis and Latif, 1981; Schloerb et al., 1950; Walike et al., 1977). The animals were deprived of food and water 16 hours prior to the initiation of this procedure. A pretreatment blood sample (3 ml) was obtained in the manner described previously (Walker et al., 1982). The animals were then immediately administered a n injection into the saphenous vein of 4 pCikg tritiated water (3HzO: Amersham Corporation, Arlington Heights, IL).Four 3ml samples of blood were drawn from these animals a t 30 minutes, 1, 2, and 3 hours postinjection, by which time equilibration of the tracer was completed. The blood samples were obtained without anesthesia, and the sera were separated by means of centrifugation and stored at -20" until later analysis. The fecal material and urine were collected from these animals for safe disposal and the ESTIMATION OF BODY FAT IN MONKEYS 325 animals were not returned to their outdoor obtain the counting efficiency for each unsocial groups until their urinary radioactiv- known serum sample based on its AES ratio. Total body water (TBW) was determined ity had fallen to a level < .03 pCi/ml. In order to determine the precise water according to the following calculation (Friiscontent of the serum, a 2-ml serum sample Hansen, 1957): from each female was pipetted into a small TBW = C1V1/C2, (1) glass scintillation vial. The gross weight of C1-concentration of in‘ected 3H20, the vial and serum was recorded and the V1-volume of injected3! H20, samples were then placed in a dry-ice bath in0 serum Cz-concentration of 3 ~ 2 and frozen. Once the serum samples reached a t equilibrium. a solid state, they were placed in a lyophilizer and freeze-dried. Complete lyophilization of the samples required approximately 2.5 hours, after which time the samples were As seen in Figure 1, equilibration of the removed and reweighed. The percent water tracer was completed by 30 minutes postinin the serum samples was derived from the jection; hence the equilibrium value used in ratio of dry to wet weights and ranged from all calculations represents a mean from the 87.7% to 90.1%. To determine whether the 1-, 2-, and 3-hour samples. Computation of Total Body Fat. Lean body percent water in the samples was underestimated by using gross weight in the calcula- mass (LBM) and total body fat (TBF) were tions (i.e., including the weight of the bottle), indirectly determined using the constant, two lyophilized serum samples were rehy- 73.2, as the percent of water normally found drated and pipetted into other vials. The in fat-free mammalian tissue (Behnke, 1941; original vials were washed with ethanol and Moulton, 1923). Based on the calculation of distilled water and then allowed to air-dry. TBW, percent body fat (%BF) was computed The bottles were then weighed and the according to the following formulae (Schloerb amount of water in the serum was recalcu- et al., 1950): lated using net weights. The use .of gross (2) versus net weights yielded a mean difference 1)LBM = TBW/0.732 (3) of .02% in the estimation of the water con- 2) TBF = Body Weight (BW) - LBM (4) tent, which is well within the range of nor- 3) %BF = TBFBW mal measurement error. Since the specific Data Analysis activity of the water in the serum was used The data derived from these measurements in the calculations, the volume of serum was corrected by the percent water in serum for were analyzed using a stepwise regression analysis (Kerlinger and Pedhazur, 1973). each female. To measure the specific activity in the Simple correlational relationships between serum samples, 500 p1 of serum was placed two variables were analyzed using Pearson into microvials with 5 ml of commercial scintillation cocktail. The effect of quenching on counting efficiency was determined by pipetting increasing volumes of nitromethane (0, 2D 2 , 4 , 6 , 8 , 10,20,25 1) into scintillation vials containing 7.9 x 10 disintegrations per minute (dpm) of 3H20. Nitromethane was used because empirical studies demonstrate that this compound yields quenching similar to biological fluids and is highly stable over time. Samples were counted in a liquid scintillation spectrometer (Packard Corporation) and the automatic external standard (AES) ratio, as well as the dpm’s, were obtained for each. A quench curve plotting the AES ratio 0 34 60 120 1m versus counting efficiency was generated, yielding a linear relationship (r = .995; a = Yiiitra .068; b = 1.983). The counting efficiency for Fig. 1. The specific activity in serum of the nullipathe quench curve ranged from 4.1%(25 pl) to rous and multiparous females a t 0 minutes (preinjection) 30.0%(0 pl). This quench curve was used to and 30,60,120, and 180 minutes post 3Hz0 injection. t i 326 M.L. WALKER, S.M. SCHWARTZ, M.E. WILSON, AND P.I. MUSEY TABLE 1. Morphometric data, including X (+ SEMI and ranges, for two age classes of female rhesus monkeys Age Yr Nulliparous (n = 6) Range Multiparous (n = 7) Range 3.25 LO21 3.15-3.31 8.82 (.52) 7.25-11.25 Skinfold Crown-rump BodyTotal body Lean body thickness length weight water mass kg crn _ _ - _ kg ~mm ~ - - _ _ _ _ _ _ liters _ 1.9 45.1 4.6 3.02 4.1 (.1) (1.0) (.2) (.I61 (.2) 1.7- 2.0 43.0-48.1 3.9-5.2 2.37-3.39 3.2-4.6 6.5 50.5 7.0 4.13 5.5 (1.7) (.9) (5) (.I91 (.3) 2.0-14.2 47.8-54.3 5.4-8.7 3.56-4.92 4.9-6.7 ______ product-moment correlations (r). The two groups of subjects were treated statistically both individually and conjointly. Differences between groups were analyzed using a t test for independent measures. Data are presented as X -t SEM. RESULTS The basic morphometric data for both the nulliparous and multiparous females are given in Table 1. Determinations of subcutaneous fat depth using ultrasound were excluded, since this instrument was not sufficiently sensitive to allow for the measurement of shallow fat depth and thus did not discriminate among individuals. As can be seen from these data, the multiparous females were, as expected, heavier, taller, and fatter than nulliparous females. The relationship among these morphometric measurements and estimates of body fat for all females is shown by the correlation matrix depicted in Table 2. As can be seen from this table, all of these morphometric measures were highly interrelated (P < .05). Skinfold thickness correlated most highly with percent body fat (r2 = .go), whereas crown-rump length accounted for only 50% of the variance in predicting percent body fat (r2= 50). On the basis of these correlations, skinfold thickness appears to be the single best predictor of percent body fat, but either body weight or the Quetelet index or crownrump length alone correlates well enough for each to be a useful predictor. Interestingly, the predictability of these measures changed when the two groups of females were considered separately. As seen in Table 3, crown-rump length and percent body fat were most highly correlated (r2 = .86, P < .05) in nulliparous females. In multiparous females, however, these two variables correlated quite poorly (2 = .08, P > .05). A similar disparity in the prediction of percent body fat was found when skinfold Body fat % 7.8 (1.2) 3.6-11.1 18.3 (2.9) 9.4-32.0 TABLE 2. A correlation matrix depicting the relationship among the measurements of skinfold thickness (SF), crown-rump length (CR), body weight (B W), the Quetelet index (QI),and percent body fat (%BF) for all females BW QI %BF .81 .65 .71 .85 .87 .95 .97 .90 \ 90 \ TABLE 3. A correlation matrix depicting the relationship among the measurements of skinfold thickness (SF), crouln-rump length CCR), body weight IB W), the Quetelet index (QO, and percent body fat (%BF)' BW 2 i .38 BW CR ,531 QI .81 .98 %BF &I :;; .73 -, .29 .28 .96 .83 \ .86 %BF .45 56 .93 .16 \ 'Those values above the diagonal represent the correlations from nulliparous females whereas those helow the line are from multiparous females. thickness was used as the predictor. In multiparous females, skinfold thickness and percent body fat correlated highly (r2 = .96; P < .05),whereas in nulliparous females, these variables were nonsignificantly related (r2 = .20, P > .05). When all animals were included in a multiple regression analysis, a highly significant relationship was found amon these morphometric measures (R = .97; R8 -- .94; F p g ) = 48.14, P < .Ol), yielding the following equation: Y = + 1 . 3 ~ +1 .39X2 + (- 1 2 . 8 ) ~ ~ 14.96 where: x1 = x2 = x3 = skinfold thickness, crown-rump length, body weight. (5) 327 ESTIMATION OF BODY FAT IN MONKEYS The Quetelet index did not contribute significantly to the prediction of percent body fat and was thus statistically excluded from the regression equation. As with the simple correlations, the regression equation took a somewhat different form when the data were analyzed with the nulliparous and multipar o w females considered separately. The linear combination of height, weight, skinfold thickness, and the Quetelet index regressed against percent body fat in multiparous females yielded a correlation coefficient indicative of a strong linear relationship among these variables (R = .99; R2 = .98; F(3,3).= 43.27, P < .01). However, the degree to which each variable contributed to the equation differed from that found when the data from all subjects were included in the analysis: Y = 11.86 + 1.42x1 + . 6 3 ~ 2+ (-2.17)x3 where: x1 x2 x3 = = = (6) skinfold thickness, Quetelet index, body weight. Crown-rump length in multiparous females, however, did not contribute sufficiently to the predictability of the regression equation to satisfy the criteria for statistical inclusion. A similar stepwise regression analysis in nulliparous females indicated that a highly significant relationship exists among percent body fat and height and skinfold thickness (R = .97; R2 = .94; F(2,3)= 23.63, P < .05): Y = -52.0 + 5.65~1+ 1 . 0 5 ~ 2 (7) where: x1 = crown-rump length, x2 = skinfold thickness. However, in these females, neither the Quetelet index nor body weight contributed significantly to the equation. To illustrate the difference between the regression equations generated for the nulliparous and multiparous females, percent body fat was predicted for each group using the regression equation of the other group. The mean predicted percent body fat for the nulliparous females, using the regression equation generated from data on multiparous females, was 11.2%(k.3) as compared to the actual mean percent body fat of 7.8% (+ 1.2).The difference between the two groups in the observed versus predicted values was significant (t(5) = -2.43, P < .05). The converse analysis was also performed in which percent body fat for multiparous females was predicted from the regression equation generated from the nulliparous females. The predicted mean percent body fat was 37.7% (*lO.l) as compared to the observed mean of 18.3% (k2.9), with the difference again being statistically significant (t(6) = -2.65, P < .05). As a n internal check of the validity of the regression equations, two females were randomly selected from each group. Using data from the remaining females, new regression equations were generated. Actual body fat measures then were compared to estimates obtained from the new regression equations. The observed mean percent body fat for the two nulliparous females was 9.5% (f1.6) compared to the predicted mean of 8.5%(5.2). The observed mean percent body fat for the two multiparous females was 19.3% (k4.5) compared to the predicted mean of 23.8% (+2.9). DISCUSSION The present analyses indicate that, when taken in concert, morphometric measures can provide valid estimates of percent body fat in nonhuman primates. Although such estimates can be obtained from measures of height, weight, and abdominal subcutaneous fat depth, the utility of each may vary a s a function of the particular physical characteristics of the subject population. When younger, nulliparous and older, multiparous females were treated statistically a s a homogeneous group, the correlation of percent body fat with skinfold thickness and body weight revealed a strong linear relationship. In fact, skinfold thickness has been shown previously to correlate most highly with adiposity in pigtailed macaques (Mucucu nemestrinu), with the strongest relationship between body fat and skinfold thickness obtained from obese females (Walike et al., 1977). Our data concur with this finding, since in multiparous females, who tended to weigh more and have greater subcutaneous fat depth, the correlation between skinfold thickness and percent body fat was much higher than in smaller nulliparous females. Certain disparities in the predictive relationship among such morphometric measures as body weight, crown-rump length, and skinfold thickness have been found to occur in subjects who are at the extremes of age, height, and weight (Damon and Goldman, 1964). As can be seen from our data, the 328 M.L. WALKER, S.M. SCHWARTZ, M.E. WILSON, AND P.I. MUSEY range in skinfold thickness measures in the nulliparous females was quite small. Evidently, due to a truncation in skinfold thickness values, this measure is a very poor predictor in young females when used alone. Moreover, since the lower limit of accurate measurement of skinfold thickness is approximately 2 mm (Hansen, personal communication), young females are ill-suited to this sort of morphometric estimation of body fat. Instead, crown-rump length, which correlated most highly with percent body fat in this age group, is a better predictor of percent body fat in younger females. In multiparous, fully adult females, skinfold thickness was a n excellent predictor of percent body fat. Body weight and the Quetelet index also correlated highly with determinations of body fat. On the other hand, crown-rump length correlated very poorly with percent body fat, despite a wide range of heights in these females. Furthermore, the prediction of body fat derived from the regression equation from the inappropriate age group (i.e., predicting nulliparous values from the multiparous regression equation and vice versa) yielded widely disparate values for predicted and observed percent body fat. In contrast, when data from both nulliparous and multiparous females were used in the regression equations, estimates of body fat were similar to actual measurements. These data reemphasize the necessity to distinguish a priori the physical traits of the animals and to select measurement techniques accordingly. In addition they illustrate that when using established regression equations to generate estimates of body fat, the population from which the equations are derived must encompass the full range of physical characteristics represented by the population to be estimated. The total body water method for determining body fat seems particularly well suited for use in rhesus monkeys since the fraction of water in fat-free tissue in smaller animals has been found to be relatively constant (Pace and Rathbun, 1945; Pitts, 1956). Whatever measurement error results from total body water determinations is generally dismissed as simply reflecting irreducible variability in individual body composition (Siri, 1961), although the degree of experimental error is proportionately larger in smaller animals. In addition, determinations of body fat from total body water is more practical than other techniques such as carcass analysis, water submersion, and soft tissue roentgenograms. The high predictive accuracy between the observed and predicted body fat estimates, as generated from the regression equations, underscores the validity and general applicability of this method. In fact, our work generally concurs with the finding in humans of a linear increase in percent body fat occurring with age (Friis-Hansen, 1957). This study illustrates that indirect assessments of morphology in macaques can provide accurate estimates of body composition, circumventing the need for more invasive and laborious techniques. Furthermore, multiple morphometric measures, taken in combination, provide more accurate estimates of body fat than single measures taken alone. The method of choice is a matter requiring consideration of the developmental status of the animals. ACKNOWLEDGMENTS The authors would like to thank Dr. Barbara Hansen for her helpful advice and support. We would also like to thank Louise Wright for her editorial assistance and Thomas P. Gordon for his comments. This work was supported, in part, by NSF grant BNS13248, NIH grants RR00165, HD16305, and VA Project 1517-005. LITERATURE CITED Behnke, AR (1941) Fat content and composition of the body. Harvey Lect. 37:198-226. Billewicz, WZ, Kemsley, WFF, and Thomson, AM (1962) Indices of adiposity. Br. J. Prev. SOC. Med. 16t183-188. Brogek, J (1961)In J Brozek and A Henschel (eds): Techniques for Measuring Body Composition. Washington, D.C.: Nat. Acad. Sci. Natl. Res. Council, pp. 3-35. Brogek, J, and Kinzey, W (1960)Age changes in skinfold compressibility. J. Gerontol. 15:45-51. Bullen, BA, Quaade, F, Olesen, E, and Lund, SA (1965) Ultrasonic reflections used for measuring subcutaneous fat in humans. Hum. Biol. 37:375-384. Damon, A, and Goldman, RF (1964) Predicting fat from body measurements: Densiometric validation of ten anthropometric equations. Human Biol. 36:32-44. Durnin, JV,and Rahaman, MM (1967) The assessment of the amount of fat in the human body from measurements of skinfold thickness. Br. J. Nutr. 21;681-689. Friis-Hansen, B (1957) Changes in body water compartments during growth. Acta Paediatrica 46:l-68. Haymes, EM, Lundegren, HM, Loomis, JL, and Buskirk, ER (1976) Validity of the ultrasonic technique as a method of measuring subcutaneous adipose tissue. Ann. Human Biol. 3:245-251. Kamis, AB, and Latif, NB (1981) Turnover and total body water in macaque (Macaca fascicularis) and gibbon (Hylobates lar). Comp. Biochem. Physiol. 70Ar4546. Kerlinger, FN, and Pedhazur, E J (1973)Multiple Regres- ESTIMATION OF BODY FAT IN MONKEYS sion in Behavioral Research. New York: Holt, Rinehart and Winston, Inc. Keys, A, Fidanza, F, Karvonen, MJ, Kimura, N, and Taylor, HL (1972) Indices of relative weight and obesity. J. Chron. Dis. 25:329-343. Moulton, CR (1923) Age and chemical development in mammals. J. Biol. Chem. 57:79-97. Pace, N, and Rathbun, EN (1945) Studies on body composition. 111. The body water and chemically combined nitrogen content in relation to fat content. J. Biol. Chem. 158:685-691. Pitts, GC (1956) Body fat accumulation in the guinea pig. Am. J. Physiol. 185:41-48. Schloerb, PR, Friis-Hansen, BJ, Edelman, IS, Solomon, AK, and Moore, FD (1950) The measurement of total body water in the human subject by deuterium oxide dilution. J. Clin. Invest. 29:1296-1310. Schultz, AH (1929) The techniques of measuring the outer body of human fetuses and of primates in general. Contrib. Embryol. Carnegie Inst. Washington XX:213-257. Siri, WE (1961) Body composition from fluid spaces and 329 density: Analysis of methods. In J Brozek and A Henschel (eds): Techniques for Measuring Body Composition. Washington, DC: Nat. Acad. Sci. Natl. Res. Council, pp. 223-244. Small, MF (1981) Body fat, rank and nutritional status in a captive group of rhesus macaques. Int. J. Primatol. 291-95. Walike, BC, Goodner, CJ, Koerker, DJ, Chideckel, EW, and Kalnasv. LW (1977) Assessment of obesitv in Dietailed maca&es (~I?ucacanemestrinu). J. Med Prikatol. 6:151-162. Walker, ML, Gordon, TP, and Wilson, ME (1982) Reproductive performance in capture-acclimated female rhesus monkeys (Mucuca mulattd. J. Med. Primatol. 11:291-302. Wilmore, JH,and Behnke, Ar (1970) An anthropometric estimation of body density and lean body weight in young women. Am. J. Clin. Nutr. 27:267-274. Young, CM, and Blondin, J (1962) Estimating body weight and fatness in young women. J. Am. Dietetic Assoc. 41:452-455.