点击显示 收起
1 From the Nutrition Unit & Hepato-Gastroenterology Service (JCD), the Clinical Research and Biostatistic Unit (PMP), the Rheumatology Service (CB), and the Neurology Service (PPC), Dupuytren University Hospital, Limoges, France, and the Human Nutrition Laboratory (CB-D and BB) and Neurology Service (PC), University Hospital, Clermont-Ferrand, France.
2 Supported by the Limoges University Hospital, Limoges, France. 3 Reprints not available. Address correspondence to JC Desport, Unité de Nutrition, Service d’Hépato Gastroentérologie, CHU Dupuytren, 87042 Limoges cedex, France. E-mail: nutrition{at}unilim.fr.
ABSTRACT
Background: Amyotrophic lateral sclerosis (ALS) is a neurologic disease with an unfavorable prognosis that may be accompanied by malnutrition or overweight. Nutritional status is difficult to evaluate in these patients because of their physical limitations and the asymmetry of their disease involvement. Bioelectrical impedance analysis (BIA), which enables bedside analysis of body compartments, has not been adequately validated for use in patients with ALS.
Objective: We compared reference measures of fat-free mass (FFMa), obtained by dual-energy X-ray absorptiometry, with FFM obtained by BIA and by the skinfold-thickness technique.
Design: We measured FFMa in 32 ALS patients. Anthropometric measures included weight, height, skinfold thickness, and arm and wrist circumferences. The fat mass obtained from the skinfold-thickness measures enabled us to calculate FFM. BIA was performed by measuring the bioimpedances at 5, 50, and 100 kHz of each side of the body and from one side to the other. FFM was calculated by using the instrument’s internal software and by using 3 standard equations. The concordance between the methods was evaluated by the Bland-Altman test.
Results: Two of the 16 measured FFM values were not significantly different from FFMa. However, the risk of dispersion was too high to be acceptable in practice. An equation was then developed by using multivariate analysis, with impedance at 50 kHz. This equation was validated in a second population of 15 ALS patients and with the use of 2 successive measurements performed on 18 patients.
Conclusion: BIA is a simple technique that is valid for use in ALS patients, both for a single exam measure and for longitudinal monitoring, with the use of an adapted equation and a frequency of 50 kHz.
INTRODUCTION
Amyotrophic lateral sclerosis (ALS) is a serious disease of unknown origin that affects motor neurons. Its prevalence is estimated at between 2.5 and 6.4/100 000 habitants (1, 2), with a possible recent mortality increase in certain countries (3, 4). It can initially present in a bulbar form, marked by difficulty with swallowing and speaking, or a spinal form, with an amyotrophy of the lower or upper limbs. The disease course is rapidly downward, with a median survival of between 23 and 52 mo (5). In addition to late age at onset and initial bulbar onset, several prognostic factors have been identified, among which is malnutrition: malnourished patients have a significantly worse median survival time than do nonmalnourished patients (6). Malnutrition is caused by swallowing difficulties, which first impair oral alimentation and then later make it impossible. Other factors may also be implicated, such as denervation, psychological difficulties, digestive problems, and hypermetabolism, the origin of which is not completely elucidated (7). By contrast, in the absence of swallowing difficulties or when oral or enteral feeding is too great, excess fat mass can occur. It is thus important to evaluate body composition and its variations in patients with ALS.
In practice, the principle tools used to evaluate body composition in patients with ALS are anthropometric, with the calculation of body mass index and the measurement of weight and change in weight, triceps skinfold thickness (TSF), and arm muscle circumference (6, 8–16). It is also possible to measure 4 cutaneous skinfold thicknesses and derive body density, then fat mass (FM), and finally fat-free mass (FFM) according to the methods of Durnin and Womersley (17, 18). Body mass index, weight, and change in weight do not provide information on the body compartments and thus do not allow for the evaluation of possible losses in FFM or increases in FM. The TSF and arm muscle circumference examine FM and FFM, respectively, but the validity of these measures is arguable in the setting of ALS for several reasons: 1) abnormal fat distribution is possible in ALS, linked, for example, to a predominant involvement of the lower limbs in spinal forms of the disease (5); 2) the disease can have effects that are more marked on one side of the body than the other (5); and 3) in a nonspecific manner, TSF and arm muscle circumference have an operator-based risk of variation (19, 20). The measurement of 4 skinfold thicknesses can reduce the risk of error due to inhomogeneous disease involvement of the body, but the intraobserver risk persists.
Dual-energy X-ray absorptiometry (DXA), considered to be a reference technique, enables FM and FFM to be differentiated but requires a costly machine and the subject to remain flatly supine for 10 min, which may be difficult for ALS patients to comply with. Moreover, DXA has been used in only a few studies involving few patients (10, 13, 21) and is not available for performing total body measurements except in certain centers. On the contrary, bioelectrical impedance analysis (BIA), which explores body composition, is an inexpensive and atraumatic method that can be quickly and easily used at the patient bedside (6, 12, 22). However, this method has been incompletely validated for ALS patients (23). Additionally, the different frequencies used (50 kHz or 5 and 100 kHz) have not been compared. Thus, the aim of the present study was to compare the reference FFM obtained by DXA (FFMa) with FFM obtained by using BIA with different methods of calculation and by using skinfold-thickness measurement techniques.
SUBJECTS AND METHODS
The patients studied had ALS meeting the criteria defined by El Escorial (24). They were informed of the techniques used in the study and gave their consent to participate. The study was performed according to the responsible committee on human experimentation. All of the studies were performed on the same day for each patient. Patients were studied while in a fasting state and after having urinated if possible.
Nutritional assessment
DXA measurements of FM, lean soft tissue mass, and total body bone mineral content were performed by using a Lunar DPX-IQ 2288 (Lunar Corp, Madison, WI). The patients were scanned while in a supine position, and a total body scan took < 15 min. FFMa was calculated as weight - FM.
Anthropometric measurements were performed by the same operator and consisted of a weight measurement made with the patient in a seated position on an electronic SECA chair scale (Vogel & Halke, Hamburg, Germany) that recorded to 0.1 kg and was tared at zero before each measurement. A chair scale was used because the patients were unstable when standing upright on a standard weight scale. Both wrist and arm circumferences (at the middistance between the acromion and the olecranon with the arms straight at the side) were measured with a tape that recorded to 0.5 cm. Height was measured while the patients stood upright at the first examination with the use of a SECA height gauge (Vogel & Halke) that recorded to 0.2 cm. Additionally, Harpenden calipers (SEGA, Paris) were used according to standard methods to measure biceps, triceps, subscapular, and suprailiac skinfold thicknesses on both sides of the body. Each skinfold thickness was measured 3 times in succession, and the retained value represented an average of the measures of both sides of the body.
Bioelectric impedance was measured by using an Analycor3 instrument with surface electrodes (Spengler, Paris) according to standard methods (25). Measurements were performed at 50 kHz and in double frequency (5 and 100 kHz). The instrument provided impedance and FFM at 50 kHz (FFM50) and in double frequency (FFM5–100). The calculation equations of the instrument’s internal program are not known, because they are unpublished. FFM was also calculated with 3 different BIA equations (25–27). These are well-known equations selected because 1) they are simple and therefore easily applied in daily practice; 2) they take into account impedance at 50 kHz (Z50) or at 100 kHz (Z100), 2 of the most used frequencies in BIA; and 3) the Zillikens equation has been validated in both white and black populations (26). The Segal (S) equation uses height (H), weight (W), and Z100 as follows:
RESULTS
The first population studied included 32 ALS patients. Their principal characteristics are given in Table 1. The mean repeatability coefficients for the impedance measures were 0.67 ± 0.90% at 5 kHz, 0.48 ± 0.24% at 50 kHz, and 0.50 ± 0.28% at 100 kHz (n = 32, 5 repeated measures in 1 d). The mean repeatability coefficient for the DXA measures of FFM was 0.59 ± 0.11% (n = 12, 5 repeated measures in 1 d).
View this table:
TABLE 1 . Main characteristics of the first group of patients with amyotrophic lateral sclerosis (ALS) studied1
Of the 16 evaluations of FFM, only 2 were not significantly different from FFMa (Table 2). The other 14 were significantly different and hence unusable. For the 2 FFM evaluation methods that were not significantly different from FFMa, however, the comparison by the Bland-Altman method showed the risk of dispersion to be > 15%, indicating that none of these techniques were valid (Table 2).
View this table:
TABLE 2 . Fat-free mass (FFM) values obtained by the different methods, the difference between those values and reference fat-free mass obtained with dual-energy X-ray absorptiometry (FFMa), P values, and the risk of dispersion measured in the Bland-Altman plots1
In the face of the failure of the equations tested, specific equations were developed by using the variables linked to FFMa in univariate analysis with P values 0.25. These variables were sex, weight, height, wrist circumference (right and left), right arm circumference, H2/Z50 (mean, maximum, and minimum values), and H2/Z100 (mean, maximum, and minimum values) (P < 0.0001 for all). The values H2/Z50 and H2/Z100 were chosen for their known close relation with FFM (28–30). Reactance was not used because this added little precision to the prediction equations for total body water (28, 29). Multivariate analysis also took into account the mean TSF (P = 0.17), the mean subscapular skinfold thickness (P = 0.19), the mean suprailiac skinfold thickness (P = 0.17), the left arm circumference (P = 0.13), the manual muscular testing value (P = 0.07), and the Norris scale for limbs value (P = 0.21).
The final and best equation used Z50 and was as follows:
DISCUSSION
This study validated an equation permitting FFM to be measured in ALS patients, both during a single exam and at repeated examinations. This equation used BIA at 50 kHz, which is probably the most commonly used bioimpedance frequency worldwide, and from which one can deduce FFM. The measuring instruments are easy to handle and light enough to use at the patient bedside if needed.
The previously published equations that we tested were not validated. They were chosen for their simplicity and because they used bioimpedance at either 50 or 100 kHz. Equation 1 may have proved unsuitable because it was derived in a healthy adult population much younger than our patients (37 ± 2 compared with 64 ± 9 y) and all male. Our population had a sex ratio (male: female) of 1.13 (25). FFM is lower in women than in men and it decreases with age in both sexes (37), which may explain the overestimation (from 1.5 to 3.5 kg) of FFM in our patients with the use of this equation.
Equation 2 was initially established for a white population and then validated in a black population, both of which were healthy populations evenly comprising males and females (26). Here also, the subjects were younger (30 ± 7 and 38 ± 11 y for the white and black populations, respectively) than our study population. This equation gave results that were significantly different from the reference value when the mean or minimal H2/Z values were used. When the maximal H2/Z value was used, the equation paradoxically underestimated FFM by 0.8 kg, with a possible difference of 19.0% from actual values. It is the only equation among those we tested that integrated neither patient weight nor sex, only height and Z50, which are thus insufficient in the case of ALS patients, who are known to have perturbations of body composition. These alterations, relative to a baseline population, include either a loss of FFM associated with a loss of FM when alimentary intake is insufficient (10, 12) or a loss of FFM associated with stability in or an increase in FM when alimentary intake is sufficient or excessive (8, 13, 21).
Equation 3 gave FFM values that were consistently different from the reference value (27). It was developed in healthy subjects who were older than our patients (70 ± 5 y for the men and 68 ± 3 y for the women, compared with 63 ± 12 y for our population). This may in part explain the significant underestimation of FFM with the use of this equation (from 6.5 to 9.0 kg). With the internal algorithms of the BIA instrument, the values of FFM obtained with Z5 and Z100 differed too significantly from the reference mean to be utilized. The result was similar for the mean and minimum FFM values obtained with Z50. The maximum FFM value obtained with Z50, although not significantly different from the reference, could not be used because of the large risk of dispersion compared with the actual value (16.0%). It was not possible to analyze these results because the internal equations of the instrument’s software are unpublished.
Finally, the skinfold thickness method also could not reliably determine FFM in our patients. This may be explained in part by the known variability of the method (20) and by the fact that the zones measured by skinfold thickness may be affected by the disease, but in an asymmetric and inconsistent fashion. For example, one patient might have involvement that is more marked on one side of the body than the other, or even a predominant involvement of the lower limbs, leaving the upper portion of the body relatively unscathed (5).
BIA is a straightforward and atraumatic technique but has been infrequently used in ALS patients. Kasarskis et al (12) used an RJL Systems Inc (Detroit) instrument in 16 patients. Although the authors did not specify, it was probably a monofrequency (50 kHz) instrument. Total water and percentage FM and FFM were determined, and a tendency for FM to drop in the terminal stage of ALS was noted, more markedly in men than in women. The same tendency was noted for FFM estimated by anthropometric and biological measures performed at the same time, but the results for bioimpedance were not given. BIA was nevertheless not validated. The same team used BIA in the same manner to evaluate FFM in 18 ALS patients (22). There was a nonsignificant correlation between FFM and the bone-free arm muscle area, which is an anthropometric index closely linked with biceps and triceps muscle force and with measures of pulmonary function. The FFM obtained by BIA in this study may be arguable. Our team previously presented a validation study of BIA in which we used the internal equations of the Analycor3 instrument with impedances at 5 and 100 kHz (23). Yet, this validation study involved only 16 ALS patients and did not test the results of successive examinations. We thus decided to perform the present verification study with a new group of ALS patients.
In the present study, we found that when impedances of 5 and 100 kHz are used, FFM is significantly underestimated (by 1.9–3.5 kg; P < 0.005), making these impedances unusable in current practice. For 50 kHz, the major problem is a sizeable risk of dispersion. Hence, an alternative solution is to develop a specific equation more adapted to this population, as we have done. This equation has several advantages: being known (published) and hence verifiable; not requiring specific equipment, as instruments providing impedance at 50 kHz are widely distributed; and, finally, being applicable for a single application and for longitudinal study. The equation does not integrate patient sex because this, although considered in the multivariate analyses, was found not to be significant. Paradoxically, the multivariate analysis model retained TSF, despite the limitations that we have already identified with respect to anthropometry. TSF remains useful in the patients studied; without it, the equation explained only 89.5% of the variance with a residual SD of 3.7. Kasarskis et al (22) suggested in their study that used bone-free muscle area that anthropometric measurements of upper limbs may be acceptable in ALS patients, as indeed several authors have done (8, 12, 18). However, anthropometric measurements have never really been validated.
In our longitudinal study, FFM was underestimated at T1 by 0.4 kg and was overestimated at T2 by 1.3 kg. Nevertheless, there was no significant difference between FFM obtained by our equation and FFMa at the 2 times; consequently, these differences may have been due to chance. In clinical practice, we think that differences of 0.9% at T1 and 3.3% at T2 for FFM allow the use of our equation in the follow-up of ALS patients. The only neurologic index correlated with FFMfin50 between T1 and T2 was the Norris scale for bulbar function, but there was a great difference between the percentage of variation for this index and that for FFMfin50 (29.0% and 6.8%, respectively). Moreover, the Norris scales rely on subjective criteria (6), suggesting that it may not be an appropriate tool for the FFM follow-up.
The possibility that the hydration status of the FFM in patients with ALS is abnormal cannot be eliminated; abnormal hydration status is found in other neurologic disorders, such as spinal cord injuries (38, 39). The coefficient of conversion of total body water to FFM (/0.74) might then vary, constituting an additional factor explaining, for example, the unsuitability of Equations 1, 2, and 3 or a part of the difference in the variation in FFM between T1 and T2. Additionally, the hydration status can affect the determination of FFM by anthropometry, and DXA calculations assume a constant hydration state of FFM (40), which might bring the validity of the reference method itself into question. One manner of resolving this issue would be to determine the total body water of patients by the isotope dilution method, determine the total body density by the skinfold-thickness method, and then calculate the FM by using the Siri equation for 3 body compartments (31), which integrates total body water and does not rely on the hydration status of FFM. Finally, FFM could be obtained by subtracting FM from weight. We have applied this technique in patients with spinal cord injury (41), but it is more difficult to perform in patients with ALS because of their reduced stamina for such testing and because of the complexity of the testing.
ACKNOWLEDGMENTS
We thank Valerie Newman-Chalifour for translating the manuscript.
JCD, PPC, PMP, CB-D, PC, and BB were responsible for study design; JCD, CB, and PPC were responsible for data collection; JCD and PMP were responsible for data analysis; and JCD, PMP, PPC, CB-D, PC, and BB were responsible for writing the manuscript. All the authors certify that they had no financial or personal interest, including advisory board affiliations, in any company or organization sponsoring the research.
REFERENCES