Estimation of Arm Adipose Tissue Quotient Using Segmental Bioimpedance Spectroscopy

Abstract

Background:

Breast cancer-related lymphedema (BCRL) is a chronic condition characterized by accumulation of lymph fluid that may subsequently become fibrotic with infiltration of adipose tissue. Bioimpedance spectroscopy (BIS) is the preferred method for early detection of lymphedema as it can estimate extracellular lymph fluid. This study developed a modified impedance technique that concurrently estimates both lymph accumulation and increases in adipose tissue.

Methods and Results:

BIS was used to estimate the adipose tissue volume in a cohort of healthy women (n = 171), which was found to be highly correlated (r > 0.87) with measurements of adipose tissue obtained using the reference method of dual-energy X-ray absorptiometry (DXA). In a separate cohort of women with BCRL (n = 16), adipose volumes measured by BIS and reference method, respectively, were 2452.9 ± 933.3 mL and 2109.1 ± 824 6 mL for affected arms; 1770.9 ± 747.8 mL and 1801.4 ± 775.7 mL for unaffected arms; and comparable values for a group of age-matched controls were 1862.5 ± 661.6 mL and 1657.0 ± 641.1 mL for age-matched control arms. The increase in adipose tissue in affected arms was significant irrespective of the method of measurement, p < 0.02 and p < 0.001 for BIS and DXA, respectively.

Conclusions:

An impedance method is described that can estimate increase both in lymph accumulation and adipose tissue in breast cancer-related lymphedema.

Introduction

Surgery and radiotherapy for treatment of cancer may damage a patient's lymphatic system, resulting in failure to transport lymph adequately with consequent accumulation of lymph fluid. This condition is known as lymphedema¹ and is an adverse outcome for some women following treatment for breast cancer (breast cancer-related lymphedema, BCRL).² The incidence of BCRL has been reported to be as high as 56%, although most studies suggest that incidence rates of about 20% are more typical.³

Treatment of BCRL may be facilitated by early detection of BCRL.⁴ Currently, the most common method used for early detection is measurement of limb size or swelling. Swelling is assessed typically as an increase in limb circumference, measured with a tape measure, or increase in limb volume, either computed from limb circumference and length measurements or directly by water displacement or perometry.^5,6 During the past two decades, bioimpedance spectroscopy (BIS) is being increasingly used for the early detection of BCRL due to its ability to quantify extracellular fluid (ECF), of which lymph is a principal component.^7–9 However, in chronic BCRL, apart from lymph accumulation, excess adipose tissue may also occur.^10–13 Clinically, increased adipose tissue may lead to a change in approach to treatment, for example, the use of liposuction.¹⁴ It would therefore be ideal if bioimpedance could provide a simple technique to simultaneously estimate arm adiposity as well as arm lymph volume.

Bioimpedance is commonly used in nutrition as an inexpensive, quick, and noninvasive means of estimating overall body fat mass (FM).^15,16 Technically, bioimpedance actually measures the fat-free mass (FFM) of the body from which FM is then estimated indirectly by difference with body weight (BW).¹⁷ While impedance can be similarly used to predict limb FFM, it is not generally possible to estimate limb FM in this way since the overall weight of a limb is not known.¹⁸ However, a novel method for predicting total body FM from segmental bioimpedance, circumference, and length measurements has been proposed.¹⁹ This approach calculates an adipose tissue quotient (Q), defined as the ratio of adipose tissue volume to adipose volume plus FFM volume, for each measured segment and does not require knowledge of total tissue mass. Although Q was determined for the limbs, the method was validated for prediction of whole body FM only rather than that of individual body segments. Subsequently, Ward et al.²⁰ adapted the method to estimate the volume of adipose tissue of the arm and validated the method against limb adipose tissue mass determined by the reference method, dual-energy X-ray absorptiometry (DXA). A disadvantage of that approach, however, was that it required the measurement of impedance, length, and circumferences of multiple sublimb segments, a time-consuming process. By comparison, impedance, when used to assess BCRL, requires a single measurement of arm impedance and no anthropometric measurements.

The aim of the present study was to modify the method of Ward et al.²⁰ for assessment of arm adiposity to use only a single whole arm impedance measurement.

Materials and Methods

Participants

Control data for 170 women, aged 18–73.5 years, were obtained from a database of body composition and impedance data maintained at the University of Queensland. Participants were all healthy adults whose characteristics are presented in Table 1. Five different research studies provided suitable participant data. Data were obtained between 2005 and 2012. All necessary data were collected according to identical protocols, but not necessarily with the same instrumentation (as described below). All studies were approved by the Medical Research Committee of the University of Queensland.

Table 1.

Participant Characteristics

	Controls		BCRL
	All	Similar age subset	Affected arm	Unaffected arm
Number	170	20	16	16
Arm affected	Not applicable	Not applicable	9 Right, 7 left	7 Right, 9 left
Dominance	Not known	Not known	14 Right, 2 left	2 Right, 14 left
Age (years)	39.5 ± 10.2 (18.3–73.5)^a	56.9 ± 7.4 (32.0–73.5)	60.4 ± 11.1 (32.0–79.0)^b
Height (cm)	165.4 ± 6.3 (147.7–187.4)	162.1 ± 5.6 (150.0–170.0)	161.9 ± 6.1 (150.0–172.5)
Weight (kg)	73.9 ± 14.5 (44.1–117.0)	69.8 ± 12.6 (50.0–101.1)	72.9 ± 16.5 (50.5–117.7)
Body–mass index (kg/m²)	27.0 ± 4.8 (17.8–40.6)	26.5 ± 4.6 (18.6–37.0)	27.6 ± 4.9 (20.8–40.8)
DXA arm fat (g)¹	1227.3 ± 622.5 (188.5–3362.1)^a	1260.1 ± 473.0 (442.3–2439.7)^a	1596.4 ± 624.1 (566.5–2982.8)^b,c	1409.7 ± 500.7 (785.1–2890.1)^b,d
DXA arm lean (g)¹	2021.9 ± 415.5 (1144.7–4007.1)	1821.1 ± 271.9 (1274.1–2358.0Q)	1718.6 ± 558.4 (915.3–2996.7)^c	1477.3 ± 416.8 (896.2–2292.5)^d
DXA arm BMC (g)¹	136.2 ± 22.7 (82.3–262.5)^a	119.7 ± 14.7 (94.7–144.3)^b	119.8 ± 26.4 (79.8–168.7)^b	111.5 ± 17.4 (73.8–143.0)^b
R₀ (ohm)	340.6 ± 48.2 (227.0–480.7)^a	358.9 ± 44.9 (271.0–434.4)^a	272.0 ± 63.6 (165.6–383.4)^b,c	321.9 ± 55.7 (185.0–408.1)^d
R₀ ratio (unaffected:affected)			1.214 ± 0.222 (1.000–1.740)
R_∞ (ohm)	247.1 ± 32.9 (177–360.9)^a	254.0 ± 33.1 (212.0–318.8)^a	209.4 ± 40.4 (119.8–279.5)^b,c	232.3 ± 38.7 (147.0–289.7)^d

Values with different superscripts are significantly different; a versus b, p < 0.001, independent t test, controls versus BCRL; c versus d, p < 0.02 paired sample t test, affected versus unaffected.

Values for wrist to axilla right arm.

BCRL, breast cancer-related lymphedema; BMC, bone mineral content; DXA, dual-energy X-ray absorptiometry.

Data for 16 women who had clinically ascribed unilateral BCRL were available for reanalysis from a previous study,²⁰ which had ethical approval from the University of Sydney and the University of Queensland Human Research Ethics Committees.

Anthropometric measurements

BW, including lightweight indoor clothing, was measured to the nearest 0.1 kg using digital scales. Height was measured to the nearest 0.1 cm using a wall-mounted stadiometer.

Impedance measurements

Segmental tetrapolar BIS measurements of the arms were obtained with either an SFB3 or SFB7 impedance spectrometer (ImpediMed Ltd., Brisbane, Australia) for control participants or SFB7 only for BCRL participants. All measurements were obtained according to the protocol described previously for the assessment of whole arm according to the principle of equipotentials.²¹ Briefly, the impedance of the whole arm, wrist to axilla, was determined using skin surface electrodes located at the wrists and ankles. Measurements were available for the right arm only for controls and both arms for BCRL participants.

Both impedance devices measured impedance at 256 logarithmically distributed frequencies from 3 kHz to 1 MHz. Impedance data were analyzed as described previously and according to the Cole–Cole model using Bioimp, v4.10.0, software provided by the manufacturer. Resistances at zero (R₀) and at infinite frequency (R_∞) were obtained. R₀ is the resistance of ECF and is used as an index of lymph accumulation. R_∞ is the resistance of total tissue water and is used to determine tissue FFM.²² As previous studies comparing the performance of the SFB3 and SFB7 instruments observed a small difference (1.6%) in R values when measuring the same subject, all SFB3 values were converted to their SFB7 equivalent values using the conversion factors provide by Ward.²³

DXA measurements

Reference composition of the right arm was determined with two different DXA scanners; a Lunar DPX-IQ (GE Healthcare, Chicago, IL) or an Hologic QDR 2000 (Hologic, Inc., Bedford, MA) depending upon the study. Whole arm composition (lean, bone mineral, and fat), including the hand, was obtained using the manufacturers' region of interest software. It is well recognized that small differences, in body composition of ∼3%, are observed between DXA instruments from different manufacturers or models within manufacturers.²⁴ All Hologic data were therefore converted to their Lunar equivalent values using previously determined conversion factors.²⁵ Composition of both arms for the BCRL participants was obtained using an Hologic QDR Explorer (Hologic Inc.) using manufacturer's region of interest software that includes the hand to the axilla.

To obtain composition data for the region from the wrist to the axilla to correspond to the region measured by impedance, DXA data were corrected for mean hand composition obtained from other studies²⁶ (unpublished data). DXA reports tissue composition data, including that of fat, bone mineral content (BMC), and lean tissue, as mass of tissue. The respective mass values were converted to their volume equivalents as described previously²⁰ assuming a density of 0.87 kg/L for fat, 4.55 kg/L for BMC, and 1.05 kg/L for lean. Since the conductive volume of the arm comprises lean and adipose tissues, which are distinct from fat, fat volume was converted to adipose tissue wet volume assuming a mean hydration of 13%.²⁷ Bone mineral is considered anhydrous and nonconducting in the model (described below).

Theory

Electrical model of the arm

According to the model proposed by Biggs et al.,¹⁹ the arm may be considered as an equivalent cylinder comprising three concentric cylinders. The central core represents bone with negligible conductivity equivalent to infinite resistance and therefore not contributing to the measured impedance. The central bone core is surrounded by a cylindrical annulus comprising principally skeletal muscle with high conductivity. This is then surrounded by the outer cylinder representing adipose tissue with much lower conductivity, owing to its lower water content than skeletal muscle, but nonetheless contributing to the measured impedance. The impedance of the adipose compartment is modeled as \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} { R_ { adipose } } = { \rho _ { adipose } } \ \frac { L } { { { A_ { adipose } } } } \tag { 1 } \end{align*} \end{document}

where R_adipose is the resistance (ohm) of the adipose tissue compartment; ρ_adipose is the resistivity (ohm·cm) of the conductive element of the adipose compartment; L is the length (cm); and A is the cross-sectional area (cm²) of the cylindrical model of the adipose tissue compartment.

The analogous model for the arm muscle compartment is \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} { R_ { muscle } } = { \rho _ { muscle } } \ \frac { L } { { { A_ { muscle } } } } \tag { 2 } \end{align*} \end{document}

Determination of adipose tissue quotient

The adipose tissue quotient (Q) of the arm is defined as \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} Q = { \frac { { V_ { adipose } } } { \left( { { V_ { adipose } } + { V_ { muscle } } } \right) } } \tag { 3 } \end{align*} \end{document}

where V_adipose and V_muscle are the volumes of adipose tissue and muscle, respectively, determined by DXA. Since lean tissue in the arm is principally muscle, V_muscle was assumed to be equivalent to the DXA-measured lean volume. If the Equations (1 –3) are combined using the method proposed by Biggs et al.,¹⁹ the adipose tissue quotient of the arm may also be calculated from impedance measurements as \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} Q = 100 { \frac { { \rho _ { adipose } } } { { \rho _ { muscle } } - { \rho _ { adipose } } } } \left( { { \frac { { \rho _ { muscle } } { L_ { Arm } } } { { R_ \infty } { A_ { con } } } } - 1 } \right) \tag { 4 } \end{align*} \end{document}

where ρ_adipose is the resistivity of adipose tissue fat; ρ_muscle is the resistivity of muscle; L_Arm is the length of the arm; A_con is the cross-sectional area of the conductive tissue; and R_∞ is the extrapolated resistance at infinite frequency.

To complete the equation, a number of assumptions were made. The resistivity at infinite frequency (ρ_adipose) of the adipose compartment fat was assumed to be 34 ohm·m from the compendia of reference values.^28–30 The resistivity of the appendicular skeletal muscle (ρ_muscle) was calculated since only values of the total body lean resistivity were found in the literature and these may not be applicable to the limbs. ρ_muscle was calculated from Equation (4) using BIS measurements and DXA-derived Q values [Eq. (3)] for a randomly selected subset (50%) of control participants. The value obtained for the resistivity of arm muscle (ρ_muscle) was 1.71 ± 0.29 ohm·m. The length of the arm, from the wrist to the axilla, was assumed to be 31.2% of height based on reference body proportionality data as used previously.³¹

Since the intention was to develop a method that required no additional measurements than those obtained during conventional BIS measurements for BCRL, circumferences were calculated according to the relationships described by Heymsfield et al.,³² in which the circumference of the equivalent cylinder representing the arm is estimated from measurements of height, weight, and age for males and females, respectively, as \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} \begin{split} lo{g_e}Circumference = & 3.88 + 0.024 \ lo{g_{e}} \ Age \\ & + 0.62 \ lo{g_e} \left( {W / H} \right)\end{split} \tag{5} \end{align*} \end{document}

where Circumference is arm circumference (cm); W is weight (kg); H is height (cm); and Age is age in years.

For the BCRL group, the estimated circumference of the affected arm was increased to correct for the change in size due to lymph accumulation. The increase was based on increase in volume of the affected to unaffected arm as determined from the R₀ ratio.

A_con [Eq. (4)] is the cross-sectional area of the conductive tissue compartment (muscle plus adipose) and is calculated from the total arm cross-sectional area and the nonconductive compartment (bone) area. \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} {A_{con}} = {A_{total}} - {A_{noncon}} \tag{6} \end{align*} \end{document}

A_total was determined from the conventional geometric relationship: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} { A_ { total } } = { \frac { Circumferenc { e^2 } } { 4 \pi } } \tag { 7 } \end{align*} \end{document}

using arm circumference calculated by Equation (4).

The model assumes that the nonconductive bone can be represented by a single cross-sectional area, A_noncon. In reality, the upper arm contains a single bone, the humerus, whereas the lower arm contains two bones, the radius and the ulna. The average wet bone area was thus determined from the average of the cross-sectional areas of the humerus and the sum of the cross-sectional areas of the radius and ulna. The cross-sectional areas of each bone were calculated by Equation (8)¹⁹: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} {A_{bone}} = {K_a}*L_{seg}^2 \tag{8} \end{align*} \end{document}

where the coefficients, K_a = 0.03 for the upper arm and K_a = 0.02 for the lower arm, were determined from peripheral quantitative computed tomography (pQCT) images²⁰ and L_seg values for the upper and forearm were calculated from the wrist to the axilla length, assuming a ratio of upper arm to lower arm length of 0.937.³³ Since it is only the bone mineral that is nonconducting rather than skeletal wet bone, the nonconductive bone area, A_noncon, was calculated assuming a wet bone density of 1.82 g/mL³⁴ and BMC density of 4.55 g/mL.²⁰

Data analysis

The relationship between adipose tissue quotients determined from impedance measurements [Eq. (4)] and those determined directly from adipose and muscle volumes measured by DXA [Eq. (3)] was assessed by Passing and Bablok regression and Lin's concordance regression coefficient (r_c). Agreement between values was assessed using limits of agreement analysis³⁵ and paired samples t test. Comparison between groups was analyzed using group t test for independent samples. Statistical analysis was performed using Medcalc (v 17.1).

Results

The characteristics of the participants are presented in Table 1. There were no significant differences between the groups of women in height, weight, and body–mass index. The controls were, however, significantly younger than the BCRL women. To provide a better comparison group for those women, a similarly aged subset of 20 women was obtained from the total control cohort. Exact case–control age matching was not possible as this would result in an age-matched subset of only 11 women. Fifteen BCRL participants were aged between 50 and 79 years; the remaining participant was aged 32 years. Nineteen of the controls were aged 50–74 years and a single participant was 32 years old. These participants provided a subset of individuals who were not significantly different in age or physical characteristics.

Arm FM measured by the reference technique, DXA, was significantly larger (p < 0.001, Table 1) in both the affected and unaffected arms of the BCRL participants compared with the control participants, although there was no significant difference in BMI, suggesting similar levels of whole body adiposity. In contrast, DXA-measured arm lean was not significantly different between BCRL and controls. BMC was significantly higher (p < 0.001, Table 1) in the total control group than in BCRL, but not in the age-similar control subset; presumably reflecting the larger bone mass in younger subjects.³⁶ In the BCRL group, both lean and FMs were larger (Table 1) in the affected arm than the unaffected arm, irrespective of limb dominance. Although overall these differences were significant (p < 0.02), not all participants showed increases in the affected arm, with 14 of 16 participants having increased fat in the affected arm and 12 of 16 participants having increased lean tissue.

Absolute resistance values were not significantly different between the unaffected arm of the BCRL participants and the right arm of the controls. However, for the BCRL participants, both R₀ and R_∞ were significantly smaller (p < 0.02, Table 1) in the affected limb than the unaffected arm reflecting the larger ECF (and total fluid) compartment. All BCRL participants, except one, had an interlimb R₀ ratio greater than the normatively determined threshold, indicative of the presence of lymphedema accounting for dominance.³⁷ The one participant who did not meet the threshold (normal range mean + 3 standard deviation [SD]) nevertheless had clinically ascribed lymphedema.

A strong correlation (r_p = 0.874) was found for the relationship between the DXA-derived and BIS-derived adipose quotients in the control subjects (Fig. 1a) with the fitted regression line lying close to the line of identity, r_c = 0.864. Notably for the BCRL participants, although 12 of 16 Q values for affected arms and 4 of 16 Q values for the unaffected arms lay outside the 1 SD envelope around the regression line, these differences, however, were not significant when tested against either of the age-matched group of controls. The limits of agreement between the two measures of Q are presented in Figure1b. The bias between measurements was essentially zero (0.002), although the 2 SD limits of agreement were wide at ±0.16 (34.4%).

FIG. 1.

(a) Correlation between adipose tissue quotient (Q) determined by BIS and by DXA. Key: regression line; , 95% confidence limits; , line of identity; ◯, control participants; ■, BCRL participant, affected arm; □, BCRL participant, unaffected arm. (b) Limits of agreement between adipose tissue quotient (Q) determined by BIS and by DXA. Key: , mean bias; , regression line; , 2 SD limits of agreement; ◯, control participants; ■, BCRL participant, affected arm; □, BCRL participant, unaffected arm. BCRL, breast cancer-related lymphedema; BIS, bioimpedance spectroscopy; DXA, dual-energy X-ray absorptiometry; SD, standard deviation.

Calculated arm adipose tissue volume was obtained from the BIS-derived Q values and compared with the adipose tissue volumes measured by DXA for the age-matched group; these data are presented in Figure 2a. There was no significant difference between the fat volumes of the unaffected arms and the age-matched control arms either when measured by BIS or DXA (BIS: 1770.9 ± 747.8 mL and 1801.4 ± 775.7 mL and DXA: 1862.5 ± 661.6 mL and 1657.0 ± 641.1 mL for BCRL and control arms, respectively). For participants with BCRL, fat volumes were significantly larger in the affected arms than unaffected arms, irrespective of the method measurement; BIS: 2452.9 ± 933.3 mL and 1770.9 ± 747.8 mL (p < 0.001) and DXA: 2109.1 ± 824 6 mL and 1862.5 ± 661.6 mL (p < 0.020), respectively.

FIG. 2.

(a) Adipose tissue volume computed from BIS-derived Q values compared with DXA-derived values in control, age-matched controls, and BCRL subjects. (b) Lean tissue volume computed from BIS-derived Q values compared with DXA-derived values in control, age-matched controls, and BCRL subjects. Key , 25th to 75th percentiles; , mean; , 95% confidence interval; ◯ ●, outside value (1.5 × the interquartile range); ■ □, far out value (3 × the interquartile range). Line style indicates participant group: , affected arm, BCRL participants; , unaffected arm, BCRL participants; , age-matched controls.

Similarly, for the BCRL participants, there were significant increases in the volume of lean tissue in the affected arm compared with the unaffected arm, irrespective of the method of measurement (Fig. 2b). BIS-derived measurements were significantly larger than DXA-derived measurements for lean tissue volume, but not fat tissue volumes (Fig. 2a, b).

Discussion

While it has been previously determined that arm adiposity can be determined using segmental bioimpedance,²⁰ this method required additional circumference measurements, making it time-consuming. Our results show that arm adipose tissue can be estimated without arm circumference measurements, but rather just using the measurements of segmental bioimpedance, height, and weight. We found that the proposed method was highly correlated with the reference method of DXA and exhibited an extremely small bias. Although precision at an individual level was low with large limits of agreement, the method was able to determine increases in adipose and lean volumes in the affected arms compared with the unaffected arms of BCRL participants in agreement with similar observations using DXA. Absolute volumes were, however, significantly different between the BIS- and DXA-derived values of lean tissue volume. Consequently, the BIS method should be most appropriately viewed as a classification or screening method for tissue composition rather than providing an accurate measurement of tissue composition per se. The observed increase in both adipose and lean tissues in the affected arms of participants with lymphedema is in agreement with the observations of Brorson et al.³⁸ who observed a similar increase in a series of women with established BCRL refractory to conservative therapy and, more recently, with the data of Zhang et al.¹³ who found increases in both fat and lean tissue of the affected arms of 141 women with unilateral BCRL. Arm tissue composition has also been shown to be modified in BCRL by whether the affected arm is dominant or nondominant.¹⁰ The effect of dominance was not studied here since this information was not available for all participants or only data for the right arm were available.

The proposed method for adipose tissue estimation not only has a number of advantages compared with the previously described method²⁰ but also some disadvantages and limitations. The method is quick and simple to use, requiring no additional measurements that would be routinely obtained in a lymphedema clinic. Impedance measurements have a high level of acceptability by patients³⁹ and are suitable for self-measurement. Although the underlying theory of the approach may appear complex and limit the method's appeal, the calculations are arithmetically straightforward and readily accomplished in a simple spreadsheet; potentially, they could be incorporated into impedance device firmware.

A potential disadvantage is that to obtain both R₀ and R_∞, the use of a BIS device is required. This excludes the use of a single low-frequency impedance device that can be used successfully for lymphedema assessment only.⁴⁰ Nevertheless, using R_∞ is more suitable than another single high frequency, for example, 50 kHz typical of simple impedance devices used for body composition assessment, since the equivalent electrical circuit of the arms has a very high capacitive component influencing the relationship between resistivity and a measured resistance.^41,42

The study and method do have a number of limitations. It should be noted that what is actually being measured is transfer impedance. To calculate conductive volumes, the resistivity of the conductivity volume between the measurement electrodes is estimated. The values used here were literature values. Strictly, these apply only to homogeneous material. Where tissue anisotropy is present, an apparent averaged resistivity can be estimated. The main handicap to an accurate estimation of average resistivity of soft tissue of the arm is the elbow joint, which effectively delineates separate lean, fat, and bone compartments for the upper and lower arms. In the previous study by Ward et al.,²⁰ this was, in part, accounted for by measuring segments of the forearm and upper arm avoiding the elbow. However, the similar performance of the whole arm approach described here and the segmental approach of Ward et al.²⁰ suggests that the use of an average apparent resistivity for the whole arm does not incur significant additional error.

The method is highly dependent upon estimation of arm volumes, total, conductive (adipose plus lean) and nonconductive (bone), based upon relationships with simple anthropometric measurements, age, and sex. These relationships were derived in a large, predominantly Caucasian, cohort of American women.³² From the present study, it would appear that these algorithms apply equally well to an Australian population. It is unknown whether they would similarly apply to populations of different body habitus, for example, Asian populations.

Although the method for derivation of adipose tissue quotients, Q, was validated in a large sample of women, 170 participants, its application to BCRL was only in a relatively small cohort of 16 subjects. This latter sample size was, however, similar to that in the study of arm adiposity using DXA by Brorson et al.,³⁸ but far less than the more recent study by Zhang et al.,¹³ in which arm composition was determined by DXA in 141 participants with lymphedema. Ideally, the present observations need to be confirmed in a larger number of women with BCRL at all stages. Data for control participants were for the right arm only and presumed to be, for the majority of participants, the dominant arm. For the BCRL participants, where dominance was known, two were left dominant. Since dominance increases overall arm size by approximately 2%–4%,⁶ not being able to account for dominance in the control cohort will have confounded comparison between control and BCRL data to a small degree. However, the primary purpose of the control participants was to demonstrate the validity of the method for determining adipose tissue quotients from impedance data for application to BCRL participants, rather than direct comparison of arm tissue composition between the two populations. The method, as described here, was applied only to those with unilateral arm lymphedema; the method could, however, be equally applied to those with bilateral arm lymphedema or lower limb lymphedema. In bilateral lymphedema, tissue composition would need to be compared with reference data for healthy controls rather than an unaffected contralateral limb; in such cases, normative reference data would need to be dominant limb specific.

In conclusion, the possibility of assessing tissue composition of arms affected by lymphedema with a simple measurement of height, weight, and impedance is attractive. It could provide useful additional information to the clinician, for example, distinguishing between early stage, predominantly fluid accumulation lymphedema, and later stage lymphedema where fat accumulation may occur and be amenable to treatment with liposuction. Since the method evaluates tissue adipose content irrespective of its origins, it also has the potential for the assessment of frank lipedema. Further research is required to confirm this possibility.

Footnotes

Author Disclosure Statement

T.E. is an employee of ImpediMed Ltd., manufacturer of impedance devices. L.C.W. consults for ImpediMed Ltd. ImpediMed Ltd. had no involvement in the concept, design, or execution of this study or in the preparation of the manuscript. All other authors have no competing financial interests.

References

Lawenda

, Mondry

, Johnstone

PAS

. Lymphedema: A primer on the identification and management of a chronic condition in oncologic treatment. CA Cancer J Clin, 2009; 59:8–24.

Morrell

, Halyard

, Schild

, Ali

, Gunderson

, Pockaj

. Breast cancer-related lymphedema. Mayo Clin Proc, 2005; 80:1480–1484.

Disipio

, Rye

, Newman

, Hayes

. Incidence of unilateral arm lymphoedema after breast cancer: A systematic review and meta-analysis. Lancet Oncol, 2017; 14:500–515.

Shah

, Arthur

, Wazer

, Khan

, Ridner

, Vicini

. The impact of early detection and intervention of breast cancer-related lymphedema: A systematic review. Cancer Med, 2016; 5:1154–1162.

Shah

, Badiyan

, Khwaja

, Dyk

, Arthur

, Lawenda

, Finkelstein

, Vicini

. Breast cancer related lymphedema: A review of recent developments. Androl Gynecol Curr Res, 2013; doi: 10.4172/2327–4360.1000102.

Ward

, Czerniec

, Kilbreath

. Operational equivalence of bioimpedance indices and perometry for the assessment of unilateral arm lymphedema. Lymphat Res Biol, 2009; 7:81–85.

Ward

, Bunce

, Cornish

, Mirolo

, Thomas

, Jones

. Multi-frequency bioelectrical impedance augments the diagnosis and management of lymphoedema in post-mastectomy patients. Eur J Clin Invest, 1992; 22:751–754.

Cornish

, Chapman

, Hirst

, Mirolo

, Bunce

, Ward

, Thomas

. Early diagnosis of lymphedema using multiple frequency bioimpedance. Lymphology, 2001; 34:2–11.

Ward

. Bioelectrical impedance analysis: Proven utility in lymphedema risk assessment and therapeutic monitoring. Lymphat Res Biol, 2006; 4:51–56.

10.

Dylke

, Ward

, Meerkin

, Nery

, Kilbreath

. Tissue composition changes and secondary lymphedema. Lymphat Res Biol, 2013; 11:211–218.

11.

Brorson

. Adipose tissue in lymphedema: The ignorance of adipose tissue in lymphedema. Lymphology, 2004; 37:175–177.

12.

Ohlin

, Olsson

, Nilsson

, Brorson

, Ohlin

, Olsson

, Nilsson

, Brorson

, Ohlin

, Olsson

, Nilsson

. Adipose tissue dominates chronic arm lymphedema following breast cancer: An analysis using volume rendered CT images. Lymphat Res Biol, 2006; 4:199–210.

13.

Zhang

, Brown

, Paskett

, Zemel

, Cheville

, Schmitz

. Changes in arm tissue composition with slowly progressive weight-lifting among women with breast cancer-related lymphedema. Breast Cancer Res Treat, 2017; 164:79–88.

14.

Brorson

, Ohlin

, Svensson

. The facts about liposuction as a treatment for lymphoedema. J Lymphoedema, 2008; 3:38–47.

15.

Pichard

, Kyle

, Bracco

, Slosman

, Morabia

, Schutz

. Reference values of fat-free and fat masses by bioelectrical impedance analysis in 3393 healthy subjects. Nutrition, 2000; 16:245–254.

16.

Kyle

, Bosaeus

, De Lorenzo

, Deurenberg

, Elia

, Gómez

, Heitmann

, Kent-Smith

, Melchior

, Pirlich

, Scharfetter

, Schols

AMWJ

, Pichard

. Bioelectrical impedance analysis—Part II: Utilization in clinical practice. Clin Nutr, 2004; 23:1430–1453.

17.

Wan

, Ward

, Halim

, Gow

, Ho

, Briody

, Leung

, Cowell

, Garnett

. Bioelectrical impedance analysis to estimate body composition, and change in adiposity, in overweight and obese adolescents: Comparison with dual-energy x-ray absorptiometry. BMC Pediatr, 2014; 14:1–10.

18.

Ward

. Segmental bioelectrical impedance analysis: An update. Curr Opin Clin Nutr Metab Care, 2012; 15:424–429.

19.

Biggs

, Cha

, Horch

. Electrical resistivity of the upper arm and leg yields good estimates of whole body fat. Physiol Meas, 2001; 22:365–376.

20.

Ward

, Essex

, Gaw

, Czerniec

, Dylke

, Abell

, Kilbreath

. Estimation of limb adiposity by bioimpedance spectroscopy in lymphoedema. J Phys Conf Ser, 2013; 12062:4072.

21.

Cornish

BHH

, Jacobs

, Thomas

BJJ

, Ward

LCC

. Optimizing electrode sites for segmental bioimpedance measurements. Physiol Meas, 1999; 20:241–250.

22.

Ward

, Isenring

, Dyer

, Kagawa

, Essex

. Resistivity coefficients for body composition analysis using bioimpedance spectroscopy: Effects of body dominance and mixture theory algorithm. Physiol Meas, 2015; 36:1529–1549.

23.

Ward

. Inter-instrument comparison of bioimpedance spectroscopic analysers. Open Med Dev J, 2009; 1:3–10.

24.

Plank

. Dual-energy X-ray absorptiometry and body composition. Curr Opin Clin Nutr Metab Care, 2005; 8:305–309.

25.

Ward

, Rudorfer

, Isenring

, Kagawa

. Body composition assessment by dual energy X-ray absorptiometry (DXA): An inter-machine comparison. Australas Med J, 2010; 3:951.

26.

Ward

, Dylke

, Kilbreath

. Measurement of hand volume by bioelectrical impedance spectroscopy. Lymphat Res Biol, 2012; 10:81–86.

27.

Thomas

. The chemical composition of adipose tissue of man and mice. Q J Appl Physiol Cogn Med Sci, 1962; 47:179–188.

28.

Gabriel

, Lau

, Gabriel

. The dielectric properties of biological tissues: II. Measurements in the frequency range 10 Hz to 20 GHz. Phys Med Biol, 1996; 41:2251–2269.

29.

Geddes

, Baker

. The specific resistance of biological material—A compendium of data for the biomedical engineer and physiologist. Med Biol Eng, 1967; 5:271–293.

30.

Report

. AL/OE-TR-1996-0037, Occupational and environmental health directorate, Radiofrequency Radiation Division, Brooks Air Force Base, Texas (USA), June 1996.

31.

Ward

(2015) Bioelectrical impedance spectroscopy for the assessment of lymphoedema: Principles and practice. Chap 12. In Lymphedema: Presentation, Diagnosis, and Treatment, Editors: Arin

. Greene, Håkan Brorson, Sumner A. Slavin, Springer Publishing.

32.

Heymsfield

, Martin-Nguyen

, Fong

, Gallagher

, Pietrobelli

. Body circumferences: Clinical implications emerging from a new geometric model. Nutr Metab (Lond), 2008; 5:1–17.

33.

de Leva

. Adjustments to Zatsiorsky-Seluyanov's segment inertia parameters. J Biomech, 1996; 29:1223–1230.

34.

Nazarian

, Von Stechow

, Zurakowski

, Müller

, Snyder

. Bone volume fraction explains the variation in strength and stiffness of cancellous bone affected by metastatic cancer and osteoporosis. Calcif Tissue Int, 2008; 83:368–379.

35.

Bland

, Altman

. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet, 1986; 327:307–310.

36.

Teegarden

, Proulx

, Martin

, Zhao

, McCabe

, Lyle

, Peacock

, Slemenda

, Johnston

, Weaver

. Peak bone mass in young women. J Bone Miner Res, 1995; 10:711–715.

37.

Ward

, Dylke

, Czerniec

, Isenring

, Kilbreath

SLL

. Confirmation of the reference impedance ratios used for assessment of breast cancer-related lymphedema by bioelectrical impedance spectroscopy. Lymphat Res Biol, 2011; 9:47–51.

38.

Brorson

, Ohlin

, Olsson

, Karlsson

. Breast cancer-related chronic arm lymphedema is associated with excess adipose and muscle tissue. Lymphat Res Biol, 2009; 7:3–10.

39.

Ridner

, Bonner

, Doersam

, Rhoten

B a

, Schultze

, Dietrich

. Bioelectrical impedance self-measurement protocol development and daily variation between healthy volunteers and breast cancer survivors with lymphedema. Lymphat Res Biol, 2014; 12:2–9.

40.

York

, Ward

, Czerniec

, Lee

, Refshauge

, Kilbreath

. Single frequency versus bioimpedance spectroscopy for the assessment of lymphedema. Breast Cancer Res Treat, 2009; 117:177–182.

41.

Buendia

, Seoane

, Bosaeus

, Gil-Pita

, Johannsson

, Ellegård

, Lindecrantz

. Robustness study of the different immittance spectra and frequency ranges in bioimpedance spectroscopy analysis for assessment of total body composition. Physiol Meas, 2014; 35:1373–1395.

42.

Buendía Lopez

. Improvements in bioimpedance spectroscopy data analysis: Artefact correction, cole parameters, and body fluid estimation. Stockholm: KTH—Royal Institute of Technology; 2013.