special communication

Transcription

1 special communication Determining bone and total body mineral content from body density and bioelectrical response spectroscopy STEVEN F. SICONOLFI, 1 RANDAL J. GRETEBECK, 2 WILLIAM W. WONG, 3 SHERIL S. MOORE, 4 AND JOHN H. GILBERT III 5 1 Neurosciences Neuromuscular Laboratory, SD3/Space and Life Sciences Research Laboratories, National Aeronautics and Space Administration Johnson Space Center, Houston 77058; 3 Stable Isotope Program, United States Department of Agriculture/Agricultural Research Service Children s Nutrition Research Center, Department of Pediatrics, Baylor College of Medicine, Houston 77030; 4 University Space Research Association, Houston 77058; 5 Aerospace Consultant Enterprises, Houston, Texas 77059; and 2 Department of Foods and Nutrition, Purdue University, West Lafayette, Indiana Siconolfi, Steven F., Randal J. Gretebeck, William W. Wong, Sheril S. Moore, and John H. Gilbert III. Determining bone and total body mineral content from body density and bioelectrical response spectroscopy. J. Appl. Physiol. 85(4): , We hypothesized that one could assess total body mineral (TBM) and bone mineral content (BMC) from measurements of body density and bioelectrical response spectroscopy (BRS)-determined total body water by using a three-compartment (3C) model. We compared TBM and BMC computed from measurements of water ( 2 H 2 O dilution or BRS) and body density (underwater weighing) with [4-compartment (4C)] and without (3C) mineral (dual X-ray absorptiometry) in 15 women and 16 men. BRS used multifrequency or single-frequency estimates of water. Mean differences between the 3C and 4C models ranged from 6.1 to 2.2%. Correlations between models were Standard errors of the estimate of % were within the range of those previously reported, i.e., %. Use of BRS did not significantly decrease the strength of the correlations between the models. A significant mean difference (only in women) was found only with 3C single-frequency BRS estimates of TBM and BMC. We concluded that investigators can assess TBM and BMC 3C multifrequency BRS estimates in men and women. multicompartment models; body composition; underwater weighing; osteoporosis One possible solution uses bioelectrical response spectroscopy (BRS) to assess total body water. BRS comprises the impedance, phase-angle, capacitance, and inductance responses of the human body to a multifrequency input current. A common problem observed with BRS is that the body s impedance shifts as fluid compartments exchange water, such as that observed when a subject moves from a standing to a supine position. The water shift between compartments may take up to 30 min to stabilize. Consequently, estimates of total body water were only valid for the amount of time the subject maintained the position used in the original research study. Siconolfi et al. (18) reported a new approach that eliminated this problem. They showed BRS assessment of total body water to be valid after 0 and 40 min of supine rest. This new BRS, combined with body density, would make remote monitoring of TBM and BMC feasible during spaceflight or in other research studies, such as field evaluation of BMC. However, we must know the effect of using water assessed by BRS on the validity of the Siconolfi et al. (17) model. This study evaluated the validity of the Siconolfi et al. estimates of TBM and BMC from measurements of body density (D b ) and water by using BRS. We evaluated the validity for all subjects (n 31) and subjects by gender (men n 16 and women n 15). SPACEFLIGHT INDUCES physiological adaptations that include decreases in bone mineral content (BMC) (15). Monitoring of BMC with bone densitometers is not possible during spaceflight because of the size and power requirements of present instrumentation. Siconolfi et al. (17) reported that one could assess total body mineral [TBM; 82.4% of which is bone (2)] from measurements of body density and water. During spaceflight, assessing body density is possible because it is the ratio of body mass (available since Skylab) to body volume (portable air-displacement volumometry) (16). The problem is the assessment of total body water on a real-time basis. METHODS Subjects. Thirty-one subjects (16 men and 15 women) participated in this study. All subjects had passed an Air Force Class III physical or its equivalent and were briefed on the study. After all questions about the study were answered, subjects signed an approved informed consent statement. Table 1 lists subject characteristics. Total body water by dilution. This method has been described in detail elsewhere (17). Briefly, subjects reported to the Johnson Space Center in the morning, after an overnight fast, for collection of baseline urine samples. They then drank an oral dose of water containing4gof99.8 atom percent 2 H 2 O (Icon Services, Summit, NJ) diluted to 100 g total volume with tap water. Urine samples collected 3, 4, and 5 h after

2 TBM AND BMC FROM BRS AND BODY DENSITY 1579 Table 1. Subject characteristics Parameter Age, yr Body mass, kg Body density, g/ml Body water, kg BMC, kg %Fat from 4C model administration of the dose were stored frozen in cryogenically stable tubes at 20 C until analysis. We prepared samples according to the procedures of Wong et al. (21) for 2 H: 1 H isotope-ratio measurements by gas-isotope-ratio mass spectrometry. The precision in our laboratory for water is 1% (on the basis of the difference between 4- and 5-h samples from 50 subjects). Total body water by BRS. The full BRS model is described in detail elsewhere (18). Briefly, we recorded BRS from a Hewlett-Packard model 4284A Precision LCR Meter through electrodes placed on the hand, wrist, ankle, and foot at standard locations, before the subject assumed the supine position. BRS uses an average input current of 250 µa (at frequencies of 5, 50, 67, 85, 100, 150, 166, 200, 250, and 300 khz). Subjects reclined in a supine position, and a computer recorded the BRS immediately. We used the Siconolfi et al. (18) parallel electric circuit model to determine water. We regressed (3rd-order least squares regression) each subject s impedance and resistance values on frequency. The mean ( SD) of the correlations for the individualized regressions was The resistance of the total circuit (RT) was the resistance at the frequency where impedance changed by only 1% with a frequency increase of 25 khz. We used the 1% limit because it is an industry standard for high-precision resistors (12). Our laboratory (4) reported that this approach was better than the statistically derived values of the Cole-Cole method. RT estimated water by using the following equation (18) Water (kg) Ht 2 /RT BM model R 0.987; SE of estimate (SEE) 1.26 kg, where Ht is the subject s height (cm), RT is the resistance ( ) of the circuit, and BM is body mass (kg). We also evaluated other BRS estimates of water on the basis of single-frequency responses. The gender-specific equations by Kushner and Schoeller (7) estimated water from Ht 2 /R (resistance at 50 khz) and BM by using the following gender-specific equations Water (kg) (Ht 2 /R) BM men: model R 0.988; SEE kg, and Values are means SD. n 31 Subjects; BMC, bone mineral content; 4C, 4-compartment model. Water (kg) (Ht 2 /R) BM women: model R 0.975; SEE kg. Body density. This protocol is described in a previously published study (17). Briefly, we determined the subject s mass on a beam balance scale and volume by underwater weighing (19). In our laboratory we used a harness and snorkel system that maintains the subject at 2.5 cm below the water surface during the weighing procedures. This reduces the effects of the water acting on the thorax. Water is thought to affect the subject s ability to reach the same residual volume observed outside the tank (3). We found no difference between residual volumes measured in subjects in this position in the water and those measured in a separate group of subjects before weighing. Therefore, we chose not to measure the residual volume during the weighing procedures in this study. The precision ( SD of the difference between 2 measurements) in our laboratory for body density is g/ml (50 subjects were weighed twice). BMC. We assessed BMC with a Hologic QDR 2000 whole body densitometer that employs the technique of dual-energy X-ray absorptiometry (DEXA). The test subject removed all metal (i.e., jewelry) or clothes containing metal before the scan. Subjects wore medical scrubs or other suitable clothing. The technician performing the scans positioned the subject in the supine position on the densitometer for each scan and instructed the subject to remain still during the scan. The Hologic scanned the subject s whole body in the pencil-beam mode. The standard analysis software provided by Hologic (version 5.64) analyzed the scans and reported BMC. Within our laboratory, the precision of the whole body scan is 0.6% (21 subjects were scanned 3 times over 5 days). The same individual performed all scans and scan analyses to avoid interoperator sources of precision error. Many studies report that DEXA measurements of BMC systematically underestimate in vivo neutron activation or chemical analysis values (5, 6, 8, 10, 13, 14, 20). Sabin et al. (14) examined the BMC of three lumbar spines (L 2 L 4 ) from 12 cadavers with DEXA and by ashing at 250 C for 60 h, followed by an additional 24 h of ashing at 500 C. They reported that DEXA systematically underestimated (14%) the ashed BMC. Ho et al. (6) also reported a systematic underestimation of BMC by DEXA. They reported the Hologic BMC to be 5.2% lower than individually ashed (n 31) lumbar spines (11 cadavers), and the regression had a SEE of 8.9%. Sabin et al. suggested that the difference between their results and the Ho et al. results were due to the different temperatures used for ashing the bone. Sabin et al. ashed eight additional vertebrae at the same temperature as did Ho et al. and found a result ( 6.7%) similar to that of Ho et al. Sabin et al. suggested that the higher temperatures used by Ho et al. may have led to chemical changes in the hydroxapatite that account for the smaller mineral content in the ashed vertebrae. Louis et al. (10) reported that neutron activation analysis also underestimated bone mineral ash (3.9%). This difference between the osseous mineral and ashed mineral was similar to that reported by dez et al. (4.3%) (11). Louis et al. also reported that DEXA was 2.2% lower than neutron activation analysis. This yielded a net 6% underestimation of DEXA BMC for the Louis et al. data. Tothill et al. (20) compared BMC from a phantom measured on Hologic, Norland, and Lunar instruments. These authors found Hologic to be 6.7% lower than the phantom (provided by Norland). They also compared BMC of humans across instruments. Hologic and Lunar BMCs were well correlated (r 0.974, Hologic 1.006*Lunar 201.2) but had significantly different means. This suggests that a correction factor should be used when the accuracy of BMC is compared with ashing or neutron activation analyses, when investigators use different bone densitometers. Two of the cadaver studies (6, 10) scanned individual lumbar spines. The BMC of these individual vertebrae are

3 1580 TBM AND BMC FROM BRS AND BODY DENSITY small, g. Lapillone et al. (8) examined the accuracy of Hologic estimates of BMC in small ( 10 g mineral) bone tablets (Ossopan) and reported that the Hologic BMC was % lower than expected. This high degree of variability makes the extrapolation to whole humans difficult. Sabin et al. (14) used the L 2 L 4 values as one measurement to avoid this error. Lapillone et al. noted that, as the BMC increased above 10 g, the error decreased from 10.9 to 3.5%. This suggests that the results of Sabin et al. could be useful as a correction factor. Heymsfield et al. (5) and Pierson et al. (13) compared in vivo neutron activation with Lunar DEXA-determined BMC. We can compare the accuracy of DEXA with this instrument after correcting the Lunar BMC to an equivalent Hologic BMC by using the regression of Tothill et al. (20). Comparing in vivo neutron activation estimate of osseous mineral by Heymsfield et al. to their DEXA data, we found that Lunar DEXA was underestimated by 3.7%. Adjusting their mean Lunar data to an equivalent Hologic response increased the underestimation to 10.8%. Lunar-determined BMC overestimated the in vivo neutron activation value of Pierson et al. by 3.7%. Adjusting their mean Lunar data to an equivalent Hologic response yielded an underestimation of 12.5%. The ashing and neutron activation studies show a range of underestimation of BMC by Hologic DEXA of %. The Heymsfield et al., Pierson et al., and Sabin et al. (14) studies produce a composite correction factor of 11.9% (adjusted for sample size). We applied each correction factor plus the adjusted mean factor to the BMC from the three-compartment (3C) dilution model (3C D ). A repeated-measures ANOVA between the DEXA BMC and corrected 3C D estimates (n 31) showed that only the Ho et al. (6) and Louis et al. (10) correction factors produced significant differences from the Hologic BMC (Table 2). The adjusted mean correction to the three 3C estimates [3C D, multifrequency BRS (3C MB ), and single-frequency BRS (3C SB )] were not significantly different ( 0.3, 1.4, and 3.0%, respectively) from the Hologic BMC for our subjects (Table 2). Therefore, we chose to use the 11.9% correction factor. TBM. Brozek et al. (2) reported that osseous mineral was 82.4% of the TBM content. We followed the suggestion by Lohman (9) and Heymsfield et al. (5) that TBM could be obtained by dividing the corrected BMC by C model assessment of TBM and BMC. Body density and total body water (with 3C D,3C MB,or3C SB ) estimated percent body fat with the following equation (9) %Fat D b w where w is total body water expressed as a fractional component of body mass. This estimate, in combination with body mass, yielded body fat. Fat-free mass (FFM) was the difference between body mass and fat mass. The density of the FFM was computed by using (17) D FFM (1 %fat)/[(1/d b ) (%fat/0.9007)] We used the equation by Siconolfi et al. (17) to determine mineral FFM (m FFM ) m FFM w FFM D FFM Multiplying m FFM by FFM resulted in TBM. BMC was 82.4% of TBM. Statistical analyses. Good validity was present when 1) mean differences between the four-compartment (4C) and 3C values were not significant (repeated-measures ANOVA with Scheffé s post hoc analyses); 2) there was a strong Pearson product-moment correlation; 3) there was a low SEE [expressed as a percentage of the 4C mean (SEE%)]; and 4) there were nonsignificant correlations between the residuals and the averaged values (1), indicating no systematic reason for differences between the methods. Validity was evaluated for all subjects (n 31) and for subjects by gender subgroups. RESULTS The means and SDs of the TBM and BMC for 4C and 3C models are presented in Table 3. Only the validity results for BMC are presented in Table 4. The TBM statistical data were not included in Table 4 because it is a function of BMC and the mean differences, correlations, SEE%, and residual correlations (1) were identical to the BMC data. A key result was the lack of a significant decrease in Pearson product-moment correlations by substituting 3C MB (r 0.90) or 3C SB (r 0.91) estimates of total body water for dilution (r 0.91) for all subjects. Male and female subgroups (Table 4) had similar results. The correlations of residuals (1) Table 2. Effects of literature correction factors on % of 3C estimates of BMC Reference Reference Method Correction Factor r* SEE% of Correction Factor % 3C D 3C MB 3C SB Ho et al. (6) Ashing 5.2% % 7.3% 6.1% 4.4% Louis et al. (10) Ashing 6.0% % 6.4% 5.3% 3.0% Sabin et al. (14) Ashing 14.0% % 2.7% 3.7% 5.3% Heymsfield et al. (5) IVNA 10.5% % 1.0% 0.1% 1.8% Pierson et al. (13) IVNA 12.5% % 1.0% 2.0% 3.6% Adjusted mean 11.9% 0.3% 1.4% 3.0% Uncorrected Hologic 13.9% 12.7% 10.7% %, Percent mean difference; % of 3-compartment model (3C) estimates of BMC (3C Hologic)/Hologic; D, MB, and SB: with dilution and multifrequency and single-frequency bioelectrical response spectroscopy, respectively. * Correlated between dual-energy X-ray absorptiometry (DEXA) BMC and reference method with standard error of the estimate (SEE): SEE expressed as a percentage of the mean reference method (SEE%). Mean adjusted for sample size from Heymsfield et al. (5), Pierson et al. (13), and Sabin et al. (14). In vivo neutron activation analysis (IVNA) adjusted from Lunar BMC to Hologic BMC by using Tothill et al. (19) regression. P 0.05 when compared with Hologic BMC.

4 TBM AND BMC FROM BRS AND BODY DENSITY 1581 Table 3. TBM and BMC assessments from 4C and 3C models in male, female, and all subjects ( 0.08 to 0.36) were not significant, suggesting no systematic over- or underestimation of BMC (Table 4) for all, male, and female subjects. The substitution of the 3C SB model estimates of total body water did produce significantly (P 0.05) lower BMC values ( 6.1%) only for female subjects. The single-frequency BRS model estimates of total body water had excellent correlations (all 0.95; men 0.92; women 0.96) with dilution values but were significantly (P 0.05) larger for all (3.9%), male (4.0%), and female (3.8%) subjects. The multifrequency BRS model estimates of total body water had both excellent correlations (all 0.96; men 0.93; women 0.97) and no significant mean differences for all (1.1%), male (1.0%), and female (1.3%) subjects. SEE% for both methods were similar ( %). DISCUSSION All Subjects, kg, kg, kg TBM 4C DEXA 3, , , C D 3, , , C MB 3, , , C SB 3, , , * BMC 4C DEXA 2, , , C D 2, , , C MB 2, , , C SB 2, , , * Values are means SD. TBM, total body mineral, 4C DEXA, 4-compartment model from DEXA. *P 0.05 compared with 4C DEXA. Table 4. Validity of 3C estimates of BMC in male, female, and all subjects % * vs. 4C DEXA r SEE% Bland-Altman r All Subjects 3C D 0.3% C MB 1.4% C SB 3.0% C D 2.2% C MB 1.7% C SB 0.6% C D 3.6% C MB 5.4% C SB 6.1% *% is 100 (3C 4C DEXA )/4C DEXA based on means. All correlations were significantly different from 0. P 0.05 compared with 4C DEXA. The effects of the single and multifrequency BRS estimates of total body water on the validity of 3C MB and 3C SB determined TBM and BMC were similar. However, the single frequency BRS model estimate of total body water underestimated the 3C SB determined TBM and BMC for the 15 female subjects in the 3C SB model. This underestimation was due to the significantly larger (3.8%) value for water from single frequency BRS compared with the isotopic dilution in our female subjects. The increase in estimated total body water increases the relative percentage of water in the FFM because FFM mass has not changed. This increase in the percentage of water in the FFM is at the expense of the percentage available for mineral and protein. The decrease in the percentage of mineral in the FFM yields smaller values for TBM and BMC. For example, if a person with an actual FFM of 60 kg (consisting of 43.5 kg of water and 16.5 kg of mineral and protein) had an estimated water content of 45.2 kg ( 4% larger than actual), the estimated residual mass decreases by 10% (14.8 kg). Assuming the percent decrease is proportionally ( 27% mineral and 73% protein based on mean female data from Table 1) distributed between mineral and proteins, the reported mineral value would be 2.7% lower than the actual mineral value. The difference between the 3C D and 3C SB models for the female subjects was 2.5%. This study also evaluated the validity of the 3C methods in male subjects. The 3C estimates of BMC were not significantly larger than were the Hologic BMC. The validity for men was equivalent for the three 3C BRS models, despite a 4% overestimation of water with the 3C SB method. This may have been due to the nonsignificant overestimation of the 3C D, which was decreased by the nonsignificant expansion of water. High correlations, small or nonsignificant mean differences, and no systematic error (evident by no significant residual correlations) are only part of the evaluation. The SEE of the 3C methods also must be examined. We expressed the SEEs as a percentage of the mean so that we could compare them to SEEs reported in the literature. Our SEEs ranged from 8.5 to 9.3%. This range of accuracy was within the range reported in the literature (4.9 13%) when DEXA BMC is compared with either ashing (of lumbar spines) or to in vivo neutron activation analysis of humans. Given the error associated with estimating total body water [ 5.6% from multifrequency BRS and 6.6% from singlefrequency BRS (18)] and the 8.8% error in the 3C estimate of BMC (17), the propagated error of using the BRS in the 3C method would be from 10.4 to 10.9%. All of the mean differences and SEEs in this study were less than these expected errors. This suggests that the estimate of BMC from the 3C MB model would be acceptable. The combined methods of a portable air-displacement volumometer (15), calibrated scale, and the BRS instrumentation can permit the assessment of TBM and BMC ( 82.4% of TBM) during spaceflight. These methods also could be used during Earth-based research, such as epidemiologic studies of osteoporosis. These instruments could easily be used in a portable laboratory (trailer) or taken into a subject s home. The total cost of the instruments would be minimal ( $15,000) compared with the cost of bone densitometers and magnetic resonance imagers. We concluded that investigators can assess TBM and BMC from the 3C MB model in men and women.

5 1582 TBM AND BMC FROM BRS AND BODY DENSITY We thank Elisabeth Spector for work with the bone densitometer and Dr. Alan D. Moore for assistance in underwater weighing. Address for reprint requests: S. F. Siconolfi, 121 Shapero Hall/ College of Pharmacy and Allied Health Professions, Wayne State University, Detroit, MI Received 3 April 1997; accepted in final form 15 June REFERENCES 1. Bland, J. M., and D. G. Altman. Statistical methods for assessing agreement between two methods of clinical measurement. The Lancet 1: , Brozek, J., F. Grande, J. T. Anderson, and A. Kemp. Densitometric analysis of body composition: revision of some quantitative assumptions. Ann. NY Acad Sci. 110: , Gibbons, E. S., G. T. Jessup, and C. J. Bunting. Exercise physiology: effect of immersion on residual volume measurements in females. J. Sports Med. Phys. Fitness 25: , Gilbert, J. H., S. S. Suire, R. Gretebeck, W. W. Wong, and S. F. Siconolfi. Effect of frequency, circuit analysis and instrument on extracellular and total body resistance (Abstract). Med. Sci. Sports Exer. 27, Suppl.: S118, Heymsfield, S. B., S. Lichtman, R. N. Baumgartner, J. Wang, and Y. Kamen. Body composition of humans: comparison of two improved four-compartment models that differ in expense, technical complexity, and radiation exposure. Am. J. Clin. Nutri. 52: 52 58, Ho, C. P., R. W. Kim, M. B. Schaffler, and D. J. Sartoris. Accuracy of dual-energy radiographic absortiometry of the lumbar spine: cadaver study. Radiology 176: , Kushner, R. F., and D. A. Schoeller. Estimation of total body water by bioelectrical impedance analysis. Am. J. Clin. Nutr. 44: , Lapillone, A., P. M. Braillon, O. Claris, P. S. Ho, P. D. Delmas, and B. L. Salle. Use of dual-energy X-ray absorptiometry for the measurements of small quantities of mineral. Biol. Neonate. 71: , Lohman, T. G. Advances in Body Composition Assessment. Champaign, IL: Human Kinetics, 1992, p Louis, O., P. VanDen Winkel, P. Covens, A. Schoutes, and M. Osteaux. Dual-energy X-ray absorptiometry of lumbar vertebra: relative contribution of body and posterior elements and accuracy in relation with neutron activation analysis. Bone 13: , dez, J. A., A. Keys, J. R. Anderson, and F. Grande. Density of fat and bone mineral of mammaliam body. Metabolism 9: , Perry, R. H. (Editor). Engineering Manual. A Practical Reference of Data and Methods in Architectural, Chemical, Civil, Electrical, Mechanical, and Nuclear Engineering (2nd ed.). New York; McGraw-Hill, 1959, p Pierson, R. N., Jr., W. J. C. Thornton, D. P. Kotler, S. B. Heymsfield, D. A. Weber, and R. M. Ma. Bone mineral and body fat measurements by two absorptiometry systems: comparison with neutron activation analysis. Calcified Tissue International 56: 93 98, Sabin, M. A., G. M. Blake, S. M. MacLaughlin-Black, and I. Fogelman. The accuracy of volumetric bone density measurements in dual X-ray absorptiometry. Calcified Tissue International 56: , Schneider, V. S., A. D. LeBlanc, and L. C. Taggart. Bone and mineral metabolism. In: Space Physiology and Medicine, edited by A. E. Nicogossian, C. L. Huntoon, and S. L. Pool (3rd ed.). Philadelphia, PA: Lea and Febiger, 1994, p Siconolfi, S. F. Soft-Sided Air Volumometer. Washington, DC: National Aeronautics and Space Administration, (NASA patent application: MSC ) 17. Siconolfi, S. F., R. J. Gretebeck, and W. W. Wong. Assessing total body protein, mineral and bone mineral content from total body water and body density. J. Appl. Physiol. 79: , Siconolfi, S. F., R. J. Gretebeck, W. W. Wong, R. A. Pietrzyk, and S. S. Moore. Assessing total body and extracellular water from bioelectrical response spectroscopy. J. Appl. Physiol. 82: , Sinning, W. E., D. G. Dolny, K. D. Little, L. N. Cunningham, A. Racaniello, S. F. Siconolfi, and J. L. Sholes. Validity of general equations for body composition analysis in male athletes. Med. Sci. Sports Exer. 17: , Tothill, P., A. Avendell, and D. M. Reid. Precision and accuracy of measurements of whole-body bone mineral: comparison between Hologic, Lunar and Norland dual-energy X-ray absorptiometers. Br. J. Radiol. 67: , Wong, W. W., L. S. Lee, and P. D. Klein. Deuterium and oxygen-18 measurements on microliter samples of urine, plasma, saliva, and human milk. Am. J. Clin. Nutr. 45: , 1987.