Body-fat measurement in patients with acquired immunodeficiency syndrome: which method should be used?13

Transcription

1 Original Research Communications-general Body-fat measurement in patients with acquired immunodeficiency syndrome: which method should be used?13 Jack Wang, Donald P Koiler, Man Russell, Santiago Burasiero, Manolo Mazariegos, John Thornton, F Avraharn Dilmanian, and Richard N Pierson Jr ABSTRACT Malnutrition is common in patients with acquired immunodeficiency syndrome (AIDS), which distorts the chemical contents in the fat-free mass (FFM) and alters the assumptions underlying the traditional methods for calculating body-fat content so that such measurements may not be accurate. In vivo neutron-activation analysis (IVNA) measures FFM independently of the traditional assumptions, thereby providing more accurate measurements of body fat. We compared seven methods for measuring body fat in 18 male patients with AIDS: IVNA, total body water (TBW by 1-l2O dilution), total body potassium (TBK by #{176}K counting), dual-photon absorptiometry (DPA), bioelectrical impedance analysis (BIA), and two wellcalibrated anthropometric methods. FatTBW and fatdpa were not significantly different from fatlvna FatTBW gave the highest correlation with fativna and the smallest SEE of ± 1.8% (1. 1 kg). The traditional and widely available TBW and the newer DPA method provide reliable estimates of fatlvna in patients with AIDS. Am J C/in Nuir 1992;56: KEY WORDS Body composition, body fat, fat-free mass, AIDS Introduction Measurements ofbody fat and fat-free mass (FFM) have been used to evaluate nutritional status and response to nutritional support. Malnutrition is a common complication in patients with acquired immunodeficiency syndrome (AIDS). We reported on body-cell-mass depletion and changes in body composition during nutritional therapy in severely undernourished AIDS patients ( 1-3). These malnutrition-induced changes affect both the fat content and the chemical contents ofthe FFM. thereby challenging the assumptions that underlie traditional methods for quantifying fat. These assumptions are that the density of the FFM and the amounts of water and potassium in the FFM are constant (4-6). We have thus pursued a method able to provide reliable body-fat measurements in malnourished AIDS patients. The introduction of effective treatments and the potential value of body-composition methods to assess nutritional status in patients in whom weight alone is a poor predictor of nutrition, requires that accurate methods for measuring body fat be validated. Of the techniques available, bioelectnc impedance analysis (BIA) is both economical and noninvasive, and can be used at the bedside. However, BIA has only been calibrated by traditional methods (7). In vivo neutron-activation analysis (IVNA) and dual-photon absorptiometry (DPA) are the most promising of the new methods because they depend on direct measurements that are free ofthe traditional assumptions (8, 9). IVNA provides separate measurements for nitrogen (marker for protein), calcium (marker for bone mineral), and each of the relatively abundant elements in the body. FFM is the sum ofthese directly measured chemical components and the total body water (TBW) measured by the widely available 3H20 (tritiated water) tracer method (8). The DPA method measures the attenuation coefficients of soft tissue, which are functions of its relative contents of fat and lean tissue (9-11). Our recent study of intermethod comparisons for measuring body fat in normal subjects showed that the results of fat estimates are method-dependent (12). Our study goals are to investigate the relationships between seven methods for estimating body fat in patients with chronic wasting states and low body fat caused by AIDS, and to develop models for predicting IVNAmeasured body fat from other more widely available indirect methods. Subjects and methods S:thjecis The 1 8 male subjects selected for this study were a subgroup of human immunodeficiency virus (HIV)-positive individuals studied in our laboratory. Subjects had been referred for evaluation ofdiarrhea and weight loss. The disease status at the time of body-composition measurements, age, weight, and height of the subjects are listed in Table 1. Six of the seven body-cornposition measurements were performed on the same day in the From the Body Composition Unit and Gastrointestinal Division, Department of Medicine, St Luke s-roosevelt Hospital Center, College of Physicians and Surgeons. Columbia University, New York, and the Medical Department, Brookhaven National Laboratory, Upton, NY. 2 Supported by NIH grants DK37352, DK42618, A12l4l4, and FOS- TWO4231. Address reprint requests to J Wang. Body Composition Unit, St Luke s-roosevelt Hospital Center, New York, NY Received February 29, Accepted for publication July 9, Ini J ( un Nuir l992;56: Printed in USA American Society for Clinical Nutrition 963

2 964 WANG ET AL TABLE 1 Characteristics and clinical status of 18 male patients with AIDS* Subject Age Height Weight BMIt TBW TBN AIDS disease Complication Stable 1 CDI kg L kg Pncu,noersli.s carinhi pneumonia Kaposi sarcoma 3 35 I Chronic diarrhea I Cytomegalovirus retinitis Pncuinocrsiis carinii pneumonia Mi cobacteriuin aviuin - intracellulare Malabsorption Cytomegalovirus retinitis, isospora Microsporidiosis Malabsorption - I Cytomegalovirus retinitis, esophageal ulcer Microsporidiosis Chronic diarrhea Persistant lymphadenopathy Microsporidiosis Histoplasmosis Cryptosporidiosis Microsporidiosis.±SD 41±10 174±6 59.3±7.1 20±2 * TBW, total body water: TBN. total body nitrogen. t In kg/rn2. Both Composition Unit at St Luke s-roosevelt Hospital. IVNA was performed at the Brookhaven National Laboratory within 2 wk of the other six measurements. The studies were delayed until the patients clinical conditions stabilized. This was necessary for the patients to be able to travel to Brookhaven National Laboratory and undergo three IVNA procedures, each of which required > 1 h to complete, as_well as to undergo the body-composition studies at St Luke s- Roosevelt Hospital which took between 4 and 5 h to complete. In addition, control of diarrhea was necessary to avoid the possibility of contaminating the instruments. As stated in Table 1, clinical stability was achieved in 14 of the 1 8 subjects, whereas the others had stabilized partially by the time of study. Body weight was measured at both St Luke s-roosevelt Hospital and Brookhaven National Laboratory. There were no significant differences between the two weight measurements. The study was approved by the Institutional Review Board of St Luke s-roosevelt Hospital Center. All ofthe measurements were made after informed consent was obtained. i%-ieasii,enzepils Body weight ± 0.2 kg, height ± 0.5 cm, skinfold thicknesses ± 2 mm, and circumferences at sites defined by Durnin and Womersley (1 3) and Steinkamp et al (14). were measured with patients in hospital gowns. Total body potassium (TBK) by #{176}K counting, TBW by 3HHO dilution with 5% correction for hydrogen loss to the nonaqueous phase ( 1 5), BIA by RJL model 101 instrument (RJL Systems, Detroit), and total body fat by DPA with the Lunar model DPX (Lunar Radiation Corporation, Madison, WI), which uses an x-ray source, were performed as described previously (12). IVNA for total body nitrogen (TBN) by prompt gamma and total body calcium (TBCa) by delayed gamma were performed as described by Cohn et al (8). C a/cu/aiion ()fhodl -fa! Three of the seven methods (TBK, TBW, and IVNA) used in this study estimate FFM. The other four methods provide estimates offat content as percent ofbody weight. Fat in absolute mass and as percent of body weight were calculated by each method. Body-fat mass was estimated as the difference between body weight and FFM. FFMTBW is derived with the Pace and Rathbun (4) constant equal to the TBW-FFM ratio of0.732; FFMTBK is estimated from the Forbes and Lewis (5) constant of68. 1 mmole/ kg for males and from IVNA by the method ofcohn et al: FFMkg water protein bone mineral = (0.95 x TBW) (TBN X 6.25) (TBCa/0.34) (8) where water is 0.95 X TBW, protein is TBN (in kg) X 6.25, and bone mineral is TBCa (in kg)/0.34. Fat as percent ofbody weight was calculated for these three methods. The percent fat from BIA was estimated by using the resistance-height2 ratio to predict total body density measured by

3 BODY FAT IN PATIENTS WITH AIDS 965 TABLE 2 Pearson s correlation coefficients (r) between methods for estimating percent body fat in male patients with AIDS TABLE 4 Pair-wise comparisons ofaverage percent body fat by seven methods in male patients with AIDS5 BIA t DUR TBW DPA IVNA STK 0.69 S BIA. bioelectrical impedance analysis: DUR, Durnin and Womersley s anthropometi-ic method (13): TBW, total body water; DPA, dualphoton absorptiometry; IVNA, in vivo neutron-activation analysis; STK, anthropometi-ic method of Steinkamp et al (14); TBK, total body potassium. F? t NS. underwater weighing, as described by Segal et al (16). The anthropometric methods ofboth Durnin and Womersley (1 3) and Steinkamp et al ( 14) give predicted fat as percent ofbody weight. Percent body fat from DPA was measured with the beef-standards calibration method described previously (1 1). The fat mass was calculated for the above four methods. Siati.siic.s Pai r-wise comparisons using Fisher s protected least-significant-difference procedure and repeated measures ofanalysis of variance were used to test the hypothesis that all methods give the same body-fat content. Linear-regression analysis was used to derive equations for IVNA-percent fat by using each of the other methods as the independent variable. The level of significance for all statistical tests was All statistical calculations were performed using Sfl TA (Computer Resource Center, Santa Monica, CA) and SAS (SAS Institute Inc. Cary, NC) statistical software packages. Results The correlations between each pair ofmethods for estimating percent body fat and fat mass are shown in Tables 2 and 3. All TABLE 3 Pearson s correlation coefficients (r) between methods for estimating body-fat mass (kg) in male patients with AIDS BIA t DUR TBW DPA IVNA STK 0.79 Average BIA t 3.lt 3.4t 8.5t 15.5t DUR Ot l4.ot TBW t DPA t l2.4t IVNA SIt 12.lt STK 7.Ot S Least significant difference = 2.4. BIA. bioelectrical impedance analysis: DUR, Dumnin and Womersley s anthropometric method ( 13); TBW. total body water: DPA, dual-photon ahsorptiometry: IVNA, in vivo neutron-activation analysis: STK, anthropometric method of Steinkamp Ct al (14): TBK. total body potassium. a = 18. t Significantly correlated at P < methods were significantly correlated with one another, except TBK and BIA. Both IVNA and TBW showed high overall correlation with other methods; the highest correlation was between the IVNA and TBW methods. TBK had low correlation coefficients with other methods forestimation ofbody fat, measured as either absolute fat mass or as percent body fat. The average percent fat by each ofthe seven methods ranged widely from 13% (8 kg) by BIA to 29% ofbody weight ( kg) by TBK as shown in Tables 4 and 5. Fat estimates from both TBK and the anthropometric method of Steinkamp et al were significantly higher than by all other methods. The least significant (P = 0.05) difference between any two methods for percent fat was 2.4, and for fat mass was 1.5 kg. The differences among IVNA. TBW, and DPA were less than the least significant value; therefore, they were not significantly different from each other. We also studied the effect of body mass index (BMI. in kg/ m2) on the accuracy of the methods. We found that there was no relationship between BMI and the differences in body-fat mass or percent fat between any methods for these patients. Short of carcass analysis, IVNA is the most direct method available for measuring the elemental components in FFM in TABLE 5 Pair-wise comparisons ofaverage body-fat mass (kg) by seven methods in male patients with AIDS5 Average BIA 1.0 I.7t 2.Ot 2.2t 5.2t 9.21 DUR t 8.2t TBW St DPA t IVNA 3.Ot 7.Ot STK 4.Ot S BIA, bioelectrical impedance analysis; DUR, Dumnin and Womersley s anthropometric method (13): TBW, total body water; DPA, dualphoton absorptiometry; IVNA, in vivo neutron-activation analysis; STK, anthropometric method of Steinkamp et al (14); TBK, total body potassium. n = 18. t NS. S Least significant difference = 1.5. BIA. bioelectrical impedance analysis: DUR. Durnin and Womersley s anthropometric method ( I 3): TBW. total body water: DPA, dual-photon absorptiometry: IVNA. in vivo neutron-activation analysis: STK. anthropometric method of Steinkamp et al (14): TBK. total body potassium. a = I 8. 1 Significantly correlated at P < 0.05.

4 966 WANG ET AL human subjects. FFM and fat measured by this method require no assumptions and the methods are not affected by disease. Models used for each of the other six methods to predict the IVNA-measured fat are shown in Tables 6 and 7. The most successful modeling was achieved with the TBW method: the intercept was not different from zero, the slope was not different from 1, and SEE was the least, at 1.9% ( 1. 1 kg). The BIA equation shows the highest SEE and the lowest correlation coefficient, as shown in Tables 6 and 7. Discussion Weight loss is a common complication in patients with AIDS. The timing of death is related to the magnitude of body-cell depletion (2). A recent study from our laboratory shows that enteral feeding in wasted patients with AIDS is associated with an increase in TBK and intracellular water indexes ofthe bodycell mass, as well as in body-fat content. These changes are associated with an improvement in overall clinical status (3). A reliable method for estimating body composition would provide a useful index for evaluating progress in treatment. However, changes in body composition due to either disease or intervention may alter the relative amounts of chemical components in the FFM and will affect the most widely used traditional methods for estimating body fat and FFM (4-6). In a recent report we made comparisons among eight methods for estimating body-fat content in normal subjects (12). Six of the seven methods used in that paper were used for this report. The average percent fat determined by each ofthe six methods is in the same high-to-low order for normal subjects and for patients with AIDS, with the single exception that TBK gives the highest percent fat for AIDS patients whereas the anthropometric method of Steinkamp et al gives the highest percent fat for normal subjects. The magnitude ofoverestimation of fat by TBK in AIDS is much greater than in normal subjects because ofa disproportionate depletion oftbk in AIDS patients, which reduces the TBK-FFM ratio (1, 2). Thus, use ofthe Forbes and Lewis constant, TBK-FFM at 68. I mmole/kg results in overestimating the fat by underestimating the FFM (4). The TBK- FFM ratio calculated for these 1 8 patients varied from 50.3 to 66 mmole/kg. The TBK-FFM ratio in AIDS patients measured by IVNA was significantly lower than that in sex- and agematched healthy subjects who volunteered for our Rosetta body- TABLE 6 Linear-regression equations for predicting neutron-activation analysis percent body fat by other methods in male patients with AIDS5 Method Constant Slope R2 SEE P TBW STK DPA DUR TBK BIA S TBW, total body water: STK, anthropometric method of Steinkamp et ai(14); DPA, dual-photon absorptiometry: DUR, Durnin and Wornersley s anthropometric method ( 13): TBK. total body potassium: BIA. bioelectrical impedance analysis. a = I 8. TABLE 7 Li near-regression equations for predicting neutron-activation analysis body-fat mass (kg) by other methods in male patients with AIDS5 Method Constant Slope R2 SEE P TBW STK DPA DUR TBK BIA S TBW, total body water: STK, anthropometric method of Steinkamp et al (14): DPA, dual-photon absorptiometry: DUR, Durnin and Wornersley s anthropometric method ( 13): TBK, total body potassium: B1A, bioelectrical impedance analysis. ii = 18. composition research project (58.4 vs 64.8 mmole/kg, P < 0.05; DP Kotler. RN Pierson Jr. and J Wang, unpublished data 1992). The more severe the depletion in TBK, the greater the overestimate of fat will be by the TBK method. Thus TBK cannot be used for estimating fat and FFM in patient with AIDS. The hydration status in these 1 8 patients was evaluated by comparing the TBW-FFM ratio in AIDS patients (0.739 ± ;. ± SD with that in 45 healthy male subjects (0.737 ± 0.039), and the Pace and Rathbun (5) estimate of The patients were not significantly different from either the measured healthy subjects or the Pace and Rathbun constant. This is one of two reasons that explains the agreement between TBW and IVNA for estimating fat and FFM in these patients. The second reason is based on the fact that TBW is the only measured variable in the Pace and Rathbun formulation (5) and the major component of FFM (74%) by the IVNA method (8). Therefore it is not surprising to see that the widely available TBW method is the most accurate one among the tested methods for estimating the IVNA-measured fat and FFM in patients with AIDS. The high SEE from the equation by DPA was unexpected. This variation may arise because at low (< 10 cm) or high (> 25 cm) soft-tissue thicknesses, the DPA method is not yet well calibrated for fat-content measurement ( 1 7). The DPA method shows great promise for fat- and soft-tissue measurement in normal body sizes, but the effects of measurement artifacts in the relatively thin AIDS patients need to be investigated further. BIA has been widely accepted as a predictor for TBW, and results offat estimates between TBW and BIA would be expected to correlate well ( I 8). In normal subjects, this model gave the lowest fat estimates ofthe seven methods (12). In AIDS patients the fatbia was very low, consistent with the findings of overhydration ofthe extracellular water we have shown in AIDS patients ( 1 ),if the single-frequency BIA used for this study actually responds to the extracellular water rather than the TBW. The BIA model was in fact originally calibrated by hydrodensitometry ( 1 6). Our recent data show that in malnourished AIDS patients total bone mineral is better preserved than is FFM. Although absolute bone mineral mass is lower in AIDS patients than in sex- and aged-matched healthy subjects, the ratios ofbone mmeral mass to FFM and TBCa to TBK are significantly higher in AIDS patients (I 9). An increased TBM-FFM ratio is associated with increased density of FFM ( 1 1 ),which will result in underestimation ofbody fat and overestimation offfm by underwater weighing. Indeed, our BIA method was calibrated by the un-

5 BODY FAT IN PATIENTS WITH AIDS 967 derwater-weighing method, which could be another reason that BIA gave the lowest estimate for fat in the AIDS patients we studied. Determining whether BIA is a better predictor for TBW or body density in patients with AIDS will require a separate study. Anthropometric methods are most suitable for the study of normal subjects in whom a steady state of energy balance would he expected to result in consistent ratios of subcutaneous to total body fat. Steinkamp et al and Durnin and Womersly derived their equations for predicting fat from anthropometric methods based on data from normal subjects; the poor results from these ill subjects by these two methods suggest that anthropometric methods that use these formulas are not suitable in malnourished patients (13, 14). It is likely that malnutrition induces major changes in visceral fat that are not detectable by anthropometric methods. Our results show that measurements by TBW and DPA provide accurate estimates for body fat in malnourished patients with AIDS when IVNA is used as a standard. The SEEs for the TBW method are much lower than those for the other methods we studied. The ratio ofextracellular to intracellular water varies with aging and disease, directly affecting the ratios of TBK to TBW and TBK to FFM (20). In patients with AIDS, TBK response to total parenteral nutrition is related to clinical conditions (2 1): therefore, patients with gastrointestinal disorders gained more TBK than did patients with systemic diseases with similar changes in FFM and fat, suggesting a different response in TBK- FFM ratio to total parenteral nutrition between these two types of patients. Although these potassium-based measurements are useful indexes for describing the quality ofthe FFM, because of the decreasing body-potassium content as malnutrition increases or the increasing potassium content as nutritional condition improves, these cannot be used for estimating fat and FFM in these patients, especially when they receive total parenteral nutrition (20). By contrast, TBW and DPA are both valid methods for measuring fat in patients with AIDS. Table 4 serves as an aid for investigators to translate from fat values determined by any ofthe listed indirect methods into that which may be considered the most direct available measurement, IVNA. (3 We gratefully appreciate the support of Yakov Kamen in performing the IVNA measurements, and thank David A Weber for his project consultation at Brookhaven National Laboratory. References 1. Kotler DP, Wang J. Pierson RN Jr. Studies of body composition in patients with the acquired immuno-deficiency syndrome. J Clin Nutr 1985;42: Kotler DP, Tiemney AR, Wang J, Pierson RN Jr. Magnitude of bodycell-mass depletion and the timing of death from wasting in AIDS. Am J Clin Nutr 1989:50: Kotler DP, Tierney AR. Ferraro R, et al. Enteral alimentation and repletion ofbody cell mass in malnourished patients with acquired immunodeficiency syndrome. Am J Clin Nutr 199 1:53: Pace N, Rathbun EN. Studies on body composition. body water and chemically combined nitrogen content in relation to fat content. J BiolChem 1945:158: Forbes GB, Lewis AM. Total sodium. potassium and chloride in adult man. J Clin Invest 1956:6: Behnke AR. Feen BG. Welham WC. Specific gravity ofhealthy man. JAMA 1942:1 18: Lukaski HC. Methods for assessment of human body composition. Am J Clin Nutr 1987:46: Cohn SH. Vaswani AN, Yasumura S. Yuen K. Ellis KJ. Improved models for determination ofbody fat by in vivo neutron activation. Am J Clin Nutr 1984:40: Mazess RB. Peppler WW. Gibbons H. Total body composition by dual-photon ( 153Gd) absorptiometry. J Clin Nutr 1984:40: Heymsfield SB. Wang J, Funfar J, Kehayias J, Pierson RN Jr. Dual photon absorptiometry: accuracy of some mineral and soft tissue mass measurements in vivo. Am J Clin Nutr 1989:49: I I. Wang J, Heymsfield SB. Aulet M, Thornton JC. Pierson RN Jr. Body fat from body density: underwater weighing vs dual-photon absorptiometry. Am J Physiol I 989:256:E Pierson RN Jr. Wang J, Heymsfield SB, et al. Measuring body fat: calibrating the rulers. Intermethod comparisons in 389 normal Caucasian subjects. Am J Physiol l99l:261:el Durnin JGVA, Womersley J. Body fat assessed from total body density and its estimation from skinfold thickness: measurements on 481 men and women aged from 16 to 72 years. Br J Nutr 1974:32: Steinkamp RC, Cohen NL, Sin WB, Sargent W, Walsh HE. Measurement of body fat and related factors in normals-li. J Chronic Dis 1965:18: I 5. Culebras JM, Moore FD. Total body water and exchangeable hydrogen I. Theoretical calculation of nonaqueous exchangeable hydrogen in man. Am J Physiol l977:232:r Segal KR, Gutin B. Presta E. Wang J, Van Itallie TB. Estimation of human body composition by electrical impedance methods: a comparative study. J Appl Physiol 1985:58: Russell-Aulet M. Wang J. Pierson RN Jr. Soft tissue thickness (TH) affects body composition measurements by dual photon absorptiornetry (DPA). FASEB J 199 l:5:1038a(abst). 18. Van Loan M, Mayclin P. Bioelectrical impedance analysis: is it a reliable estimation of lean body mass and total body water? Hum Biol 1987:59: Pierson RN Jr. Wang J. Russell-Aulet M, Kotler DP. Bone mineral mass (TBMM) and density (TBD) measured by dual photon absorptiometry (DPA) in 38 male patients with AIDS. Am J Clin Nutr I 99 1:53:P-26A(abstr). 20. Pierson RN Jr. Wang J. The quality of the lean body mass: implications for clinical medicine, In: Ellis KJ. Yasumura S. Morgan WD, eds. in vivo body composition studies. London: Inst Phys Sci Med 1987: Kotler DP, Tierney AR, Culpepper-Morgan JA. Wang J, Pierson RN Jr. Effect of home total parenteral nutrition on body composition in patients with acquired immunodeficiency syndrome. JPEN 1990:14:454-8.