Introduction

As changes in appendicular fat and fat-free tissues are typical features of malnutrition, anthropometric measurements of limb composition have long been employed as nutritional and prognostic indicators (Heymsfield et al, 1982, 1984). Bioelectrical impedance analysis (BIA) is a simple and noninvasive technique with a high potential for the assessment of limb composition (Brown et al, 1988; Heymsfield et al, 1998; Pietrobelli et al, 1998; Fuller et al, 1999a; Nunez et al, 1999; Elia et al, 2000; Lukaski, 2000; Tagliabue et al, 2000). BIA has some theoretical advantages over anthropometry because it involves less training and is generally more reproducible (Deurenberg, 1994). The validation of anthropometry and BIA for the assessment of appendicular body composition has been traditionally performed against computed tomography (CT) and magnetic resonance imaging (MRI) (Brown et al, 1988; Fuller et al, 1999b; Elia et al, 2000). Nevertheless, dual-energy X-ray absorptiometry (DXA) gives accurate estimates of appendicular fat-free tissues as compared to CT and MRI (Fuller et al, 1999b; Visser et al, 1999; Wang et al, 1999a; Elia et al, 2000; Levine et al, 2000; Shih et al, 2000). Since DXA is less invasive and/or more readily available than CT or MRI, it has a great potential for the validation of bedside techniques such as BIA. Previous comparisons of BIA with DXA have shown that BIA gives accurate estimates of appendicular fat-free tissues in healthy and overweight subjects (Brown et al, 1988; Heymsfield et al, 1998; Pietrobelli et al, 1998; Fuller et al, 1999a; Nunez et al, 1999; Elia et al, 2000; Lukaski, 2000; Tagliabue et al, 2000). However, no data are available as yet for malnourished subjects. Besides offering a more thorough evaluation of the BIA technique, these data may have prognostic implications such as it has been shown for anthropometry (Heymsfield et al, 1982, 1984).

The present study aimed therefore at evaluating the accuracy of BIA as compared to DXA for the assessment of appendicular body composition in a sample of anorexic women.

Materials and methods

Subjects

A total of 35 anorexic women followed as outpatients at the Department of Clinical and Experimental Medicine of Federico II University (Napoli, Italy) were consecutively enrolled in the study (DSM-IV, 1994). The median of AN duration was 37 months (range: 7–180 months). All anorexic women had undergone or were undergoing psychological counseling and none of them was taking oral contraceptives at the time of the study. A number of 29 age-matched healthy women recruited among the members of the medical staff served as controls. All subjects had a stable body weight (Wt) during the month prior to the study (1 kg). Controls were measured between the 6th and 10th day of the menstrual cycle. The study protocol was approved by the Ethical Committee at Federico II University and all subjects gave informed consent.

Anthropometry

Wt and height (Ht) were measured by the same operator following the Anthropometric Standardization Reference Manual (Lohman et al, 1988). The 50th percentile of Wt for age given by the NCHS anthropometric standards was taken as the ideal body weight (Iwt) of the subjects (Frisancho, 1990). Relative weight (RWt) was then calculated as {1-[(Wt-Iwt)/Iwt]}. Body mass index (BMI) was calculated as Wt (kg)/Ht² (m). Arm length was measured as the distance between the lateral tip of the acromion and a line joining the bony prominences of radius and ulna on the dorsum of the wrist (Organ et al, 1994). Leg length was obtained by subtracting sitting height from Ht (Lohman et al, 1988). Lt_arm and Lt_leg were calculated as the mean of left and right values.

BIA

The resistance of the whole body (R), arms (R_arm) and legs (R_leg) was measured by the same operator with a 4-polar impedance-meter (BIA 101, Akern, Firenze, Italy) at a frequency of 50 kHz using the method of Cornish et al (1999). Each subject was measured in the fasting state (8 h) and after 15 min in the supine position (Deurenberg, 1994). The coefficient of variation (CV) for BIA measurements was 2.0% at all sites, as determined by repeated daily measurements of one of the subjects. R_arm and R_leg were calculated as the mean of left and right values. The resistance index (Rl) was calculated as Ht² (cm)/R () for the whole body and as Lt² (cm)/R () for the arm and leg.

DXA

DXA allows the separation of total and segmental body mass (BM) into fat mass (FM), lean tissue mass (LTM) and bone mineral content (BMC). The sum of LTM and BMC gives fat-free mass (FFM), which was the variable of interest in this study. DXA measurements were performed using a Lunar DPX-L densitometer (Lunar Corporation, Madison, WI, USA, software ver. 3.6). FFM_arm and FFM_leg were calculated as the mean of left and right values. DXA measurements were performed by the same operator. The precision of LTM and BMC assessment, as determined by repeated weekly measurements of one of the subjects, was 2.5 and 1.0%, respectively. The precision of segmental LTM and BMC assessment was 3.0 and 2.0%, respectively. The difference between BM measured by DXA and Wt measured by scale was -0.20.6 kg (means.d., corresponding to -0.41.4% of Wt). In spite of its statistical significance (P<0.05, paired t-test), this difference is of no practical relevance.

Statistical analysis

Statistical analysis was performed on a MacOS computer using the Statview 5.0.1 and SuperANOVA 1.1 software packages (SAS, Cary, NC, USA). Between-group comparisons were performed by unpaired t-tests. The adjusted determination coefficient (R²_adj), the root mean square error (RMSE) and the percent root mean square error (RMSE%=RMSE/measured value of Y) obtained from linear regression of FFM vs Rl were used to determine the accuracy of BIA. Measured and predicted values of FFM were also compared by paired t-tests. Statistical significance was set to a value of P<0.05 for all tests.

Results

The measurements of the study subjects are given in Table 1.

Table 1 - Measurements of anorexic and control women (means.d.).

Full table

Anorexics and controls had a similar age and Ht (P=NS) but Wt, RWt and BMI were significantly lower in the former (P<0.0001). FFM was lower in anorexics than controls when expressed as an absolute value (P<0.0001) but higher when standardized on Wt (P<0.0001). A lower BMC (P=0.001) contributed to the lower FFM of anorexics. A lower quantity of FFM was observed also in the arms (P<0.0001) and legs (P<0.0005) of anorexics as compared to those of controls. R (P<0.0001) and R_arm (P<0.0005) were higher in anorexics while R_leg was similar to controls (P=ns). Lt_arm and Lt_leg did not differ between anorexics and controls (P=ns).

The prediction models obtained by regressing FFM vs Rl in controls and anorexics are given in Table 2.

Table 2 - Relation between FFM and RI in controls and anorexics.

Full table

RI explained 55, 42 and 36% of the variance of FFM, FFM_arm and FFM_leg, respectively (P<0.0001). The corresponding values of RMSE% were 5, 10 and 8%. When the predictive algorithms generated on controls were applied to anorexics, they overestimated FFM (39.72.8 vs 36.83.9 kg, P<0.0001), FFM_arm (1.90.1 vs 1.60.2 kg, P<0.0001) and FFM_leg (7.50.6 vs 6.81.0 kg, P<0.0001; paired t-test). We decided therefore to develop population-specific equations for anorexics (Table 2). These equations gave satisfactory estimates of FFM, FFM_arm and FFM_leg (P=ns, paired t-test) but had higher RMSE% as compared to those developed on controls (8% vs 5% for whole body, 12% vs 10% for arm and 10% vs 8% for leg).

In controls, the residuals of the FFM–RI regression were not correlated with Wt at the whole-body (P=ns) and leg (P=ns) levels. Wt accounted, however, for 11% of the unexplained variance of FFM_arm (P<0.05). In anorexics, Wt contributed significantly to the residuals of the FFM–RI regression at all levels (FFM: R²_adj=0.34, P<0.0005; FFM_arm: R²_adj=0.42, P<0.0001; FFM_leg: R²_adj=0.13, P< 0.005).

Use of both RI and Wt as predictors of FFM_arm was not associated to any relevant improvement of RMSE% in both controls (8% vs 8%) and anorexics (12% vs 11%). An improvement was however seen when both RI and Wt were used to predict FFM (from 8 to 5%) and FFM_leg (from 10 to 7%) in anorexics (Table 3). However, RI contributed only 43% of the variance contributed by Wt (as determined by comparison of standardized regression coefficients) and the RMSE% values were not different from those obtained by regressing FFM vs Wt alone (5 and 7%, respectively). Thus, even if the contribution of RI to total FFM and FFM_leg was undoubtedly significant (P0.009), Wt alone may be a superior predictor for practical purposes.

Table 3 - Prediction of whole body and leg FFM from RI and weight in anorexics.

Full table

Discussion

BIA offers accurate estimates of total body composition in anorexics, but its accuracy for the assessment of appendicular body composition in these subjects is not known (Hannan et al, 1990; Scalfi et al, 1993, 1997; Polito et al, 1998). In the present study, population-specific equations were needed to obtain accurate estimates of total and appendicular FFM from BIA in anorexics. BIA algorithms developed on healthy subjects generally fail when applied to ill subjects, probably because of differences in the underlying body water distribution (Bedogni et al, 1996). R is highly dependent on the extra- (ECW) to intra-cellular (ICW) water ratio and it is by means of this association that BIA allows an assessment of total body water (TBW) and FFM (Deurenberg, 1994; Heymsfield and Wang, 1994). A greater variability of the ECW:ICW ratio may be responsible for the lower accuracy of BIA and for the greater contribution of Wt to the unexplained variance of total FFM and FFM_leg in anorexics. However, this hypothesis needs to be tested by directly measuring TBW and ECW and by establishing their effect on the estimate of FFM obtained with BIA.

The finding that Wt contributed to the unexplained variance of FFM in anorexics but not in controls deserves some comments. In homogeneous samples such as those studied here, Wt is generally a better predictor of TBW and FFM than Rl, while the contrary happens for heterogeneous samples (Kushner et al, 1992; Scalfi et al, 1997). The reason why Wt did contribute to the unexplained variance of FFM in anorexics but not in controls is unclear but it is possible that a frequency of 50 kHz may not be enough to obtain an accurate estimate of FFM in anorexics because of changes in their ECW: ICW ratio.

Frequencies >50 kHz may be superior for assessing limb composition by BIA because they allow a better evaluation of ICW (Deurenberg, 1994). Since the extremities are made mainly of muscles, and muscles are made by water for about 80% of their weight (Wang et al, 1999b), there appears to be a strong physiological reason why future studies of appendicular body composition should consider using frequencies >50 kHz. This may be especially true for ill subjects because subclinical water shifts occur frequently with disease (Bedogni et al, 1996, 1997). In a study of healthy subjects performed by Pietrobelli et al (1998), frequencies >100 kHz were associated with an increase in the explained variance of FFM_arm and FFM_leg. However, in a study by Tagliabue et al (2000), the opposite was observed, so that there is a clear need for further research on this topic.

This study confirms the potential of BIA for the assessment of appendicular body composition in malnourished subjects. Future studies should consider whether frequencies >50 kHz allow a better assessment of appendicular body composition in anorexics as compared to Wt alone.

References

1. Bedogni G, Bollea MR, Severi S, Trunfio O, Manzieri AM & Battistini N (1997): The prediction of total body water and extracellular water from bioelectric impedance in obese children. Eur. J. Clin. Nutr. 51, 129–133. | Article | PubMed | ChemPort |
2. Bedogni G, Polito C, Severi S, Strano CG, Manzieri AM, Alessio M, Iovene A & Battistini N (1996): Altered body water distribution in subjects with juvenile rheumatoid arthritis and its effects on the measurement of water compartments from bioelectric impedance. Eur. J. Clin. Nutr. 50, 335–339. | PubMed |
3. Brown BH, Karatzas T, Nakielny R & Clarke RG (1988): Determination of upper arm muscle and fat areas using electrical impedance measurements. Physiol. Meas. 9, 47–55.
4. Cornish BH, Jacobs A, Thomas BJ & Ward C (1999): Optimizing electrode sites for segmental bioimpedance measurements. Physiol. Meas. 20, 241–250. | Article | PubMed | ISI | ChemPort |
5. Deurenberg P (1994): International consensus conference on impedance in body composition. Age Nutr. 5, 142–145.
6. DSM-IV (1994): Diagnostic standardization manual. Washington: American Psychiatric Association.
7. Elia M, Fuller NJ, Hardingham CR, Graves M, Screaton N, Dixon AK & Ward LC (2000): Modeling leg sections by bioelectrical impedance analysis, dual-energy X-ray absorptiometry, and anthropometry: assessing segmental muscle volume using magnetic resonance imaging as a reference. Ann. NY Acad. Sci. 904, 298–305. | PubMed |
8. Frisancho A (1990): Anthropometric Standards for the Assessment of Growth and Nutritional Status. Ann Arbor: The University of Michigan Press.
9. Fuller NJ, Hardingham CR, Graves M, Screaton N, Dixon AK, Ward LC & Elia M (1999a): Predicting composition of leg sections with anthropometry and bioelectrical impedance analysis, using magnetic resonance imaging as reference. Clin. Sci. 96, 647–657. | PubMed |
10. Fuller NJ, Hardingham CR, Graves M, Screaton N, Dixon AK, Ward LC & Elia M (1999b): Assessment of limb muscle and adipose tissue by dual-energy X-ray absorptiometry using magnetic resonance imaging for comparison. Int. J. Obes. Relat. Metab. Disord. 23, 1295–1302. | Article | PubMed |
11. Hannan WJ, Cowen S, Freeman CP & Shapiro CM (1990): Evaluation of bioelectrical impedance analysis for body composition measurements in anorexia nervosa. Physiol. Meas. 11, 209–216.
12. Heymsfield SB, Gallagher D, Grammes J, Nunez C, Wang Z & Pietrobelli A (1998): Upper extremity skeletal muscle mass: potential of measurement with single frequency bioimpedance analysis. Appl. Radiat. Isot. 49, 473–474. | PubMed |
13. Heymsfield SB, Mc Manus CB, Smith J, Stevens V & Nixon DW (1982): Anthropometric assessment of muscle mass: revised equations for calculating bone-free muscle area. Am. J. Clin. Nutr. 36, 680–690. | PubMed | ISI | ChemPort |
14. Heymsfield SB, Mc Manus III C, Seitz SB, Nixon DW & Andrews JS (1984): Anthropometric assessment of adult protein-energy malnutrition. In Nutritional Assessment, RA Wright & SB Heymsfield (eds) pp 27–82. Boston: Blackwell Scientific Publications.
15. Heymsfield SB & Wang Z (1994): Bioimpedance analysis: modeling approach at the five levels of body composition and influence of ethnicity. Age Nutr. 5, 106–110.
16. Kushner RF, Schoeller DA, Fjeld CR & Danford L (1992): Is the impedance index (Ht²/R) significant in predicting total body water? Am. J. Clin. Nutr. 56, 835–839. | PubMed | ISI | ChemPort |
17. Levine JA, Abboud L, Barry M, Reed JE, Sheedy PF & Jensen MD (2000): Measuring leg muscle and fat mass in humans: comparison of CT and dual-energy X-ray absorptiometry. J. Appl. Physiol. 88, 452–456. | PubMed | ISI | ChemPort |
18. Lohman TG, Roche AF & Martorell R (1988): Anthropometric Standardization Reference Manual. Champaign: Human Kinetics.
19. Lukaski HC (2000): Assessing regional muscle mass with segmental measurements of bioelectrical impedance in subjects undergoing weight loss. Ann. NY Acad. Sci. 904, 154–158. | PubMed |
20. Nunez C, Gallagher D, Grammes J, Baumgartner RN, Ross R, Wang Z, Thornton J & Heymsfield SB (1999): Bioimpedance analysis: potential for measuring lower limb skeletal muscle mass. J. Parenter. Enteral Nutr. 23, 96–103.
21. Organ LW, Bradham B, Gore DT & Lozier SL (1994): Segmental bioelectric impedance analysis: theory and application of a new technique. J. Appl. Physiol. 77, 98–112. | PubMed | ISI | ChemPort |
22. Pietrobelli A, Morini P, Battistini N, Chiumello G, Nunez C & Heymsfield SB (1998): Appendicular skeletal muscle mass: prediction from multiple frequency segmental bioimpedance analysis. Eur. J. Clin. Nutr. 58, 507–511.
23. Polito A, Cuzzolaro M, Raguzzini A, Censi L & Ferro-Luzzi A (1998): Body composition changes in anorexia nervosa. Eur. J. Clin. Nutr. 52, 655–662. | Article | PubMed |
24. Scalfi L, Bedogni G, Marra M, Di Biase G, Caldara A, Severi S, Contaldo F & Battistini N (1997): The prediction of total body water from bioelectrical impedance in patients with anorexia nervosa. Br. J. Nutr. 78, 357–365. | PubMed | ChemPort |
25. Scalfi L, Di Biase G, Coltorti A & Contaldo F (1993): Bioimpedance analysis and resting energy expenditure in undernourished and refed anorectic patients. Eur. J. Clin. Nutr. 47, 61–67. | PubMed |
26. Shih R, Wang Z, Heo M, Wang W & Heymsfield SB (2000): Lower limb skeletal muscle mass: development of dual-energy X-ray absorptiometry prediction model. J. Appl. Physiol. 89, 1380–1386. | PubMed | ISI | ChemPort |
27. Tagliabue A, Andreoli A, Bertoli S, Pagliato E, Comelli M, Testolin G & De Lorenzo A (2000): Appendicular lean body mass. Prediction by bioelectrical impedance analysis. Ann. NY Acad. Sci. 904, 218–220.
28. Visser M, Fuerst T, Lang T, Salamone L & Harris TB (1999): Validity of fan-beam dual-energy X-ray absorptiometry for measuring fat-free mass and leg muscle mass. Health, Aging, and Body Composition Study—Dual-Energy X-ray Absorptiometry and Body Composition Working Group. J. Appl. Physiol. 87, 1513–1520. | PubMed | ISI | ChemPort |
29. Wang W, Wang Z, Faith MS, Kotler D, Shih R & Heymsfield SB (1999a): Regional skeletal muscle measurement: evaluation of new dual-energy X-ray absorptiometry model. J. Appl. Physiol. 87, 1163–1171. | PubMed | ISI | ChemPort |
30. Wang ZM, Deurenberg P, Wei W, Pietrobelli A, Baumgartner RN & Heymsfield SB (1999b): Hydration of fat-free body mass: review and critique of a classic body-composition constant. Am. J. Clin. Nutr. 69, 833–841. | PubMed | ISI | ChemPort |