Determining the Precision of Dual Energy X-Ray Absorptiometry and Bioelectric Impedance Spectroscopy in the Assessment of Breast Cancer-Related Lymphedema

Abstract

Background:

The composition of breast cancer-related lymphedema (BCRL) has been shown to evolve from the initial accumulation of fluid to the development of fibrotic lesions and abnormal fat deposition. Therefore, precise and reliable assessments of BCRL are required to develop accurate staging and management. Although dual energy x-ray absorptiometry (DXA) and bioelectric impedance spectroscopy (BIS) have been used to assess BCRL, no study has evaluated the precision of these two modalities in the same cohort.

Methods and Results:

We determined the precision of DXA and BIS in lymphedematous (LE) and nonaffected (NA) arms of 24 women with Stage II unilateral BCRL. Precision was calculated from the results of paired bilateral arm measurements obtained from DXA scans measuring fat, lean, and bone mineral masses, BIS measuring extracellular fluid (ECF) and total fluid volume, and circumferential tape measurements (CM) of the arms to calculate the anatomic volume. Precision error was expressed as the root mean square (RMS) of the coefficients of variation (%CV) and standard deviations (SD).

Results:

The precisions of DXA and BIS varied from 1.16% (DXA measurements of LE arm total volume) to 1.86% (BIS LE arm total fluid volume) and from 0.95% (DXA lean mass of NA arm) to 1.72% (DXA BMC of NA arm). Precision of CM measures of arm volume were 1.71% CV for LE arm and 2.51% CV for NA arm. The fat and lean masses of the LE arm exceeded the NA arm by about 15% (p<0.0001). ECF and total fluid volume of LE arm was 22.6% and 19% greater than the NA arm (p<0.0001), respectively.

Conclusion:

For BCRL, these findings suggest that DXA and BIS are two measurement instruments that provide acceptable levels of precision for the measurement of arm lean mass, fat mass and ECF volume, respectively.

Introduction

Breast cancer-related lymphedema (BCRL) can be mild to severe, but often presents as a chronic, progressive multi-stage condition that is initially diagnosed by swelling of the arm(s) and/or the hand(s). The swelling is typically caused by an accumulation of protein-rich interstitial fluid that has been shown to occur in approximately 25% of women following surgical removal and/or radiation of the axillary lymph vessels and nodes.¹ BCRL can cause physical and psychological distress and negatively affects work and social functioning.² Other clinical symptoms include pain and shoulder range of motion restrictions that affect over 50% of breast cancer survivors.³ It is believed that the treatment-induced damage to the axillary region inhibits the clearing of intercellular fluid by the lymphatic system to the venous system, thus exacerbating the lymphedematous condition.⁴ Over time, the accumulation of lymph and the subsequent swelling may compromise normal tissue oxygenation, as well as muscle and immune system response, predisposing the area to recurrent infections. Also associated with chronic inflammation is an excess of fibroblasts and collagen deposition, which lead to fibrotic lesions within the edematous tissues.⁵ Although not yet completely understood, impairments to the lymphatic vessels, coupled with chronic inflammation, appear to alter the role of the lymphatic system in fat metabolism, which may result in the deposition of abnormal amounts of subcutaneous fat in lymphedematous arms.⁶ This progressive development may in fact compromise and challenge the current paradigms for assessment and treatment of BCRL.

Current decongestive therapies (e.g., compressive bandaging, manual lymphatic drainage, and remedial exercises) aid in the elimination of excess fluid in the extracellular fluid (ECF) space. However, with the more advanced stages of BCRL, the standard decongestive therapies may prove to be ineffective in reducing the cumulative fat and fibrosis.^7,8 Unless other innovative forms of therapy are introduced, the advanced stages of BCRL will remain a chronic incurable condition.^9–13 Before other forms of therapy and treatment are even considered, clinicians and health care providers must be capable of discriminating and assessing the difference between the fluid and nonfluid (e.g., total arm mass, including bone) compartments. Some of the most frequently used clinical methods (e.g., circumferential measurements and volume displacement of the arm) of assessment of BCRL fail to provide the sensitivity or are unable to differentiate between tissue types, as well as fluid and nonfluid compartments. Having access to this information would assist in directing treatment plans. Techniques such as dual-energy x-ray absorptiometry (DXA) and bioelectric impedance spectroscopy (BIS) are two measurement devices that are gaining prominence and acceptance in the lymphedema research community with regards to their safety, ease of measurement, and ability to estimate soft tissue and fluid components, respectively.^13–16 There is little published information on the precision of these two instruments for edematous upper extremities, particularly when they are both derived from one cohort of women on the same day. Precision is important in that it defines the least significant change (LSC) in tissue composition and fluid levels in the affected arms of women, as their lymphedema condition changes with treatment and over time. Furthermore, different forms of treatment therapies and interventions throughout the progressive stages of BCRL would necessitate the repeated use of DXA and BIS in the same patient cohort. Thus, precision of DXA and BIS is absolutely essential in order to determine the true effectiveness of any treatment or intervention. Therefore, the aim of this observational pilot study was to determine the precision of the DXA and BIS devices and their clinical feasibility in estimating fat and lean masses and fluid volumes in lymphedematous (LE) and nonaffected (NA) arms.

Materials and Methods

Between June 2010 and November 2011, 40 eligible patients visiting the Montreal General Lymphedema Clinic were invited to participate in this study. Twenty-four women with a clinical diagnosis of unilateral Stage II lymphedema were recruited from the clinic, and written informed consent was obtained. Eligibility criteria for the study included: 1) females between 40–70 years of age, 2) BMI≤30, 3) 6 months or more post-breast cancer treatment, 4) breast cancer Stages I-III in remission, 5) no significant chronic illness (e.g., congestive heart failure), 6) with unilateral Stage II lymphedema, 7) not pregnant, and 8) no history of sport participation involving unilateral upper extremity (e.g., tennis). Staging was determined according to the International Society of Lymphology criteria.^12,17 The staging assessments were performed by the clinic's physician and physiotherapists, who are all lymphedema specialists.

All assessments took place at the McGill Nutrition and Performance Laboratory in Montreal. Participants were 57.2±7.8 years of age, 70.1±6.3 kg in weight, and 162.1±6.7 cm in height, with a BMI of 26.7±2.6 kg/m². All values are expressed as mean±SD. Body weight was measured to the nearest 0.1 kg using a digital scale (Amcells TBS Series, USA), and height was determined without shoes to the nearest 0.5 cm using a wall mounted stadiometer.

Total body DXA scans were performed with a Lunar Prodigy Advanced scanner (GE Healthcare, Madison, WI, software version 2006). Prior to each assessment session, the measurement stability of the scanner was documented using the manufacturer supplied aluminum spine phantom. From the total body scans, appendicular fat, lean, and bone mineral content of both arm segments were identified and calculated with the region of interest extending from the gleno-humeral joint to the finger tips. Each measurement session began with two consecutive total body DXA scans with the patient supine on the DXA table. After the first scan, the patient stood up and then was repositioned on the table for the second scan. Each scan took approximately 8–10 minutes. Using previously derived densities for: fat (0.9 gm/mL); lean mass (1.1 gm/mL); bone mineral (1.85 gm/mL), the measured DXA tissue weights were transformed into estimated volumes using the following equation: v=m/d where v=volume, m=mass, and d=density.^13,18

For the BIS (Model SFB7, ImpediMed Ltd., Brisbane, Australia) measurements, each participant laid supine on a non-conductive surface, with all limbs slightly abducted. The skin surface at the sites of electrode placement was thoroughly cleaned with alcohol swabs, and four disposable dual tab electrodes were accurately placed with reference to anatomical markers. One dual tab electrode was placed at each wrist next to the ulnar head extending to the dorsal surface of the hand, 1–2 cm proximal to the metacarpo-phalangeal joint of the middle finger. The remaining two dual tab electrodes were placed at the ankle joint of each leg between the medial and lateral malleoli to the dorsal surface of the foot, 1–3 cm proximal to the metatarso-phalangeal joint of the second toe. These locations were selected according to the manufacturer's recommendations. A small constant current (200 μA) was generated and passed between the electrodes spanning the body, and the voltage drop between the electrodes provided a measure of impedance. The BIS measurements of both arms and total body took approximately 8–10 minutes. Ten minutes later after standing and removal of the electrodes, the patient was repositioned on the non-conductive surface and a new set of electrodes were placed in the identical location as before. The BIS procedure was then repeated. Before each session, a system test was run as per manufacturer's recommendation. Participants were asked to fast and refrain from consuming any fluids for 4 hours prior to their appointments.

The BIS system's analysis software (Bioimp version 2.25, ImpediMed Ltd., Brisbane, Australia) fits a semicircular locus to the reactance versus resistance data at each frequency to give an estimate of resistance at a frequency of zero and infinite. For this analysis, the parameters were set as follows: a data rejection level of “none” (default setting) and an operator-chosen frequency range of 3 KHz–500 KHz was selected. This frequency range was chosen after visual inspection of the first eight data sets and the frequency range adjustments revealed that 3 KHz–500 KHz produced better curve fits of the data to the theoretical semicircular locus in all cases.

The utility of the resistance values measured by BIS lies in the theory that, at low frequencies, the cell membrane is highly resistant. Thus, the electric current travels in the extracellular fluid surrounding the cells only, whereas at high frequencies the cell membrane is less resistant, and the electric current travels in the intra- and extracellular fluid, and thus resistance estimated at a frequency of zero (R_o) represents ECF and resistance estimated at a frequency of infinity (R_∞) represents total fluid.⁹ Using the method published by Ward, Czerniec, and Kilbreath,¹⁹ the fluid volumes for arm ECF and arm total fluid were derived from the equation: V(volume)=p (arm L²/R), where p=coefficient of resistivity, R=resistance (R_o or R_∞), and L=length of arm, (where arm length was calculated as a proportion of height).

Paired bilateral arm volume measurements were obtained by serial circumferential measurements (CM) to the nearest millimeter with a retractable no-stretch soft tape measure (Juzo, Cuyahoga Falls, OH). The CMs were taken at the following five points on the arm: 1) the mid-point of the ulna styloid, 2) 10 cm proximal to the ulna styloid, 3) antecubital fossa, 4) 10 cm proximal to the antecubital fossa, and 5) level with the axilla. The volume of each segment or truncated cone was derived from the following expression: V=l (C₁²+C₁·C₂+C₂²)/37.7, where l is the length of the segment, and C₁ and C₂ are the circumferences of each end of the segment. The total volume of the arm was the summation of these segments.²⁰

Each measurement session consisting of paired measures of all three measurement techniques and lasted approximately 60 minutes. All measurements with the DXA and BIS devices and the tape measure arm circumferential volume determinations were performed by one of the authors (AN) who had had 2 years of experience in the use of all three measurement methods.

Statistical analysis

The percent coefficient of variation (%CV) and standard deviation (SD) were calculated for each paired bilateral arm measurements as follows: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland,xspace}\usepackage{amsmath,amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6} \begin{document} \begin{align*} { \rm SD} = ( d^2 / 2 ) ^{0.5} \ { \rm and} \ \% { \rm CV} = ( { \rm SD / M} ) \times 100 \end{align*} \end{document}

where d denotes the difference between the two measurements and M denotes the mean. From these calculations the precision error also called technical error of measurement (TEM) was derived from the following expressions \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland,xspace}\usepackage{amsmath,amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6} \begin{document} \begin{align*} { \rm RMS - SD} = ( { \Sigma} {\rm SD}^2 / { \rm N} ) ^{0.5} \ { \rm and} \ { \rm RMS} - \% { \rm CV} = ( \Sigma { \rm CV}^2 / { \rm N} ) ^{0.5} \end{align*} \end{document}

where RMS is the root mean square of the paired measurements and N is the number of patients.

The least significant change (LSC) is the smallest change between two measurements taken over time that must be exceeded for the change to be considered true with 95% confidence. This is expressed as 2.77 times precision; 2.77×RMS-SD and 2.77×RMS-%CV. Theoretically, it is limited to the device from which the precision was determined, as the precision may vary with different devices.²¹

Paired t-tests were used to compare differences in variables (e.g., fat mass, lean mass, ECF, bone mineral content) between the lymphademetous (LE) and nonaffected (NA) arms.

Results

Table 1 details the differences between the measurements of the LE and NA arms for the 24 women obtained from DXA and BIS. The LE arm was significantly (p<0.0001) greater than the NA arm for all the variables except the bone mineral content. The measured percent differences between LE and NA arms in this cohort of women with Stage 2 lymphedema ranged from 15% to 22.6%. To be noted is that the total arm volume derived from the fat, lean, and bone mass DXA measurements and the CM of total volume yielded a similar percentage difference between LE and NA arms. From the DXA measurements, we found the FM of the LE arm was greater than the NA arm with a mean difference of 15.7%, while the FFM of the LE arm was greater than the NA arm with a mean difference of 15.6%. As a point of reference, a recent retrospective study, using DXA to measure, involving 1240 healthy women ranging in age from 20–80 years found the FFM of the right arm was higher than the left, with a mean difference of 5% in all age groups; yet only a slight difference of FM between the left and right arms of women over 50 years of age.²² Table 1 also shows a change in the relation between LE and NA arms after their ECF volumes, as derived from the BIS, are subtracted from their lean mass volumes. There is a reduction in the percent difference from 15.6% to 8.7% between LE and NA arms for the FFM; this percent difference is still significant.

Table 1.

Dual Energy X-Ray Absorptiometry, Bioelectric Impedance Spectroscopy, and Circumferential Tape Measurement of Arm Volumes in Women (n=24) with Stage II Breast Cancer-Related Lymphedema

Variable	Lymphedema arm	Nonaffected arm.	P	% difference
DXA, fat mass (g)	1409 (298)	1214 (270)	<0.0001	15.7
DXA, lean mass (g)	2100 (390)	1813 (177)	<0.0001	15.6
DXA, BMC (g)	141 (17)	140 (17)	0.832	0.6
DXA, total arm volume (mL)	3531 (574)	3070 (385)	<0.0001	15.0
BIS, EFC (mL)	927 (165)	756 (89)	<0.0001	22.6
BIS, total fluid volume (mL)	1806 (342)	1517 (163)	<0.0001	19.0
Circumferential arm volume (mL)	2432 (386)	2135 (304)	<0.0001	13.8
Lean mass volume minus ECF(mL)	1010 (161)	929 (130)	0.048	8.7

All values are expressed as mean±standard deviation (SD). BIS, bioelectric impedance spectroscopy; BMC, bone mineral content; DXA, dual energy x-ray absorptiometry; EFC, extracellular fluid volume.

The precision of the various measurements are listed in Table 2. Precision of DXA mass measurement for the LE arm varied from 1.28% CV (BMC) to 1.49% CV (lean mass), and for the NA arm the precision ranged from 0.95% CV (lean mass) to 1.72% (BMC). The precision of BIS measurements for LE arm were 1.65 % (ECF) and 1.86% (total fluid volume). For the NA arm, BIS precision values were 1.51% (ECF) and 1.56% (total volume). Precision for the anatomic tape measurement of volume was 1.71% CV for LE arm and 2.51% CV for NA arm. From these determinations, LSC at a 95% confidence level was calculated (Table 2). With respect to DXA fat mass, for example, a measured change ≥57.3 g or a change in %CV ≥3.91 in the LE arm would indicate, with a 95% confidence level, a significant change in fat mass. The calculations of LSC at a 95% confidence level for BIS measures of LE arm were 43.7 mL, 4.57%CV (ECF), and 90.3 mL, 5.15% CV (total fluid).

Table 2.

Precision Expressed as Root Mean Square (RMS) of the Variables in the Lymphedematous and Nonaffected Arms

Variable	Lymphedema arm		Nonaffected arm
	RMS-SD (LSC)	RMS-%CV (LSC)	RMS-SD gm (LSC)	RMS-%CV (LSC)
DXA fat mass(g)	20.7 (57.3)	1.41 (3.9)	12.4 (34.3)	1.05 (2.9)
DXA lean mass (g)	30.6 (84.7)	1.49 (4.3)	17.4 (48.2)	0.95 (2.6)
DXA bone mineral content (g)	1.8 (5.0)	1.28 (2.8)	2.5 (6.9)	1.72 (4.8)
DXA arm volume (mL)	42.0 (116)	1.16 (3.2)	18.3 (50.7)	0.62 (1.7)
BIS extracellular fluid (mL)	15.8 (43.7)	1.65 (4.6)	11.5 (31.8)	1.51 (4.2)
BIS total fluid (mL)	32.6 (90.3)	1.86 (5.2)	23.2 (64.3)	1.56 (4.3)
Circumferential arm volume (mL)	44.8 (124)	1.71 (4.7)	52.3 (145)	2.51 (6.9)

LSC, least significant change for 95% confidence.

Discussion

The findings of this study show that DXA and BIS can be used to determine differences in upper limb tissue composition and upper limb volume. We found no significant difference in precision of the DXA or BIS measurements between LE arms and NA arms. Our results show the total volume derived from DXA measurements and the CM total volume measurements yielded similar percentage differences between LE and NA arms. The discrepancy in the actual volume is attributed to the fact that there were different tissue volumes being measured in all three methods. For instance, the DXA field extended from the finger tips to the gleno-humeral joint, while the circumferential volume limits were from the wrist to several centimeters below the gleno-humeral joint, and the BIS measured from the wrist to an indeterminate point between the axilla and acromion. Importantly, although all three methods are not measuring the exact same volume, this should not affect precision.

A relatively poor inter-subject (e.g., the arm position changing the shape of the arm) and inter-tester (e.g., the degree of tightness with which the tape is applied) repeatability can influence the precision error of the measurements derived from the circumferential tape technique. Furthermore, the truncated cone method that is typically used to calculate arm volume includes the inaccurate assumption that limb segments are cone shaped and does not allow for uneven skin surface. In this study, the precision results for the anatomic tape measurement of volume were 1.71% CV for LE arm and 2.51% CV for NA arm. These coefficients of variation rank the CM measurements as the least precise of the three methods analyzed in this study.

From a research or clinical perspective, with total volume as the only result obtainable from CM, this evaluation method has limited utility for the assessment or tracking of body composition in later stage BCRL. In contrast, the DXA measures give a breakdown of the tissue differences between limbs, thereby enabling the tracking of differences in fat and lean volume over time. Clinically, the values for the LSC at a 95% confidence level calculated from the measured precision errors should serve as a guide to determine a significant change between two points in time, such as when tracking the progression of this chronic condition. For example, as previously stated, a DXA measured change ≥57.3 g of FM and/or BIS measured change ≥43.7 mL of ECF within the LE arm would indicate a significant change that may require a specific treatment or change in treatment plan. Lesser degrees of confidence enable smaller changes to be considered normal; for example, the LSC for fat at a 95% level of confidence for the LE arm is 2.77×1.41%=3.9%, whereas for an 85% confidence the LSC is 2×1.41%=2.8% (Table 2).

It is a well-known caveat that different manufacturers, instrument models, and software versions, as well as different operators, are all likely to affect the precision and reproducibility of the DXA results.^23–25 It is important to note that the lean mass as measured by DXA includes not only muscle, but all the other tissue components except fat and bone mineral (e.g., water, proteins, glycogen, nonbone minerals, etc.). There is a report wherein the DXA lean mass difference between LE arm and NA arm was attributed entirely to muscle without correcting for the presence of ECF.¹⁸ Our finding of a change in the relation between LE arm and NA arm after the ECFs are subtracted from their respective lean mass volumes show a reduction in the percent difference between LE arm and NA arm lean mass from 15.6% to 8.7%. The difference is still significant and may be due primarily to a difference in muscle mass. The BIS measurements of ECF indicate that a considerable portion of that lean mass difference found in the DXA measurements can be attributed to the fluid component of the lymphedematous arms.

Even though perometry and circumferential tape measurements are still the most commonly used clinical assessment techniques to identify the appearance of LE, other more sensitive and reliable methods have been gaining recognition.^14,26,27 Over 10 years ago, Cornish et al.⁹ demonstrated the efficacy of BIS over CM in terms of its sensitivity and specificity in being a reliable measurement tool to determine the early detection of LE in 20 women up to 24 months following post-surgical procedures for breast cancer. More recently, BIS has been found to be more sensitive than perometry in determining more localized LE.¹⁴ In fact, the lower level of detection of LE utilizing BIS has been shown to be approximately 35 mL (cit); a value that is very similar to the least significance difference (43.7 mL) found in the present study. Taking these findings into account, it would appear that the BIS technique is sensitive, reliable, and has the precision to detect the onset of fluid change that would signal the early development of LE. However, for those women who progress through the various stages of LE, there may be a greater need and use for detection devices such as the DXA that we have found to be sensitive to soft tissue changes (e.g., adipose tissue) in this particular population.

The calculations used to estimate fluid volumes with the impedance and resistance data collected with the BIS incorporates the assumption of constant coefficients of resistivity for the fluids. These coefficients of resistivity were derived from a cohort of women without lymphedema. Most likely as a result of the protein rich ECF and other resistive sensitive material (e.g., fat deposits) found in lymphedematous limbs, the coefficient of resistivity would not be the same as that for nonaffected arms and thus may be a source of error in the fluid volume calculations for the lymphedematous arms derived from the BIS. More research appears warranted to determine to what extent the body composition of the arms of women with BCRL varies over time. Although our findings have shown that the BIS measurements are less precise than DXA, we wish to point out the fact that the difference in precision between the two instruments is relatively small. This should not detract from using the BIS since, unlike the DXA scanner, tracking patients with the BIS gives information about the fluid in the limbs initially and over time. As well the BIS is portable, making bedside assessments possible; the time involved in measurement is about half that of a DXA scan and the BIS system is a fraction of the cost of the DXA scanner. The clinical need for the DXA measurements lies in those cases of BCRL where the differences in fluid volume do not account for the better part of the difference in total limb volume. Tracking patients using DXA measurements would open a window to observing if the fat content responds to traditional decongestive therapy. Hence, performing assessments of BCRL with DXA and BIS in concert will yield a more complete picture of the tissue composition and may influence treatment management.

All the technical procedures in this study were conducted by one person and although this may be ideal, it does not reflect the clinical reality wherein the patient may be monitored with these procedures over time by different personnel. The inter-individual variation in precision was not an end point in this investigation.

In conclusion, DXA and BIS are two tools with good precision for research and clinical assessment of breast cancer-related lymphedema. This finding is essential for us to acquire greater insight into the changes in tissue composition occurring with the progression of BCRL and for the development of novel treatments for this chronic condition.

Footnotes

Acknowledgments

We gratefully acknowledge the support of the Canadian Foundation for Innovation (AV), Quebec Breast Cancer Foundation (AT), and Concordia University Research Fund (RK).

Approval for this study was obtained from the McGill University Health Center's Research Ethics Board. This research was carried out in accordance with the Canadian government's Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans and the Helsinki Declaration of 1975, as revised in 2008.

Author Disclosure Statement

The authors declare that they have no conflict of interest. No competing financial interests exist.

Abbreviations Used

References

Petrek

, Senie

, Peters

, Rosen

. Lymphedema in a cohort of breast carcinoma survivors 20 years after diagnosis. Cancer, 2001; 92:1368–1377.

Thomas-MacLean

, Hack

, Kwan

, Towers

, Miedema

, Tilley

. Arm mobility and disability after breast cancer: New directions for care. Oncol Nurs Forum, 2008; 35:65–71.

Lawenda

, Mondry

, Johnstone

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

Devoogdt

, Van Kampen

, Gearts

, Coremans

, Christiaens

. Different physical treatment modalities for lymphoedema developing after axillary lymph node dissection for breast cancer: A review. Eur J Obstet Gynecol Reprod Biol, 2010; 149:3–9.

Tassenoy

, De Mey

, Stadnik

et al. Histological findings compared with magnetic resonance and ultrasonographic imaging in irreversible postmastectomy lymphedema: A case study. Lymphat Res Biol, 2009; 7:145–151.

Alitalo

. The lymphatic vasculature in disease. Nat Med, 2011; 17:1371–8.

Brorson

, Svensson

. Liposuction combined with controlled compression therapy reduces arm lymphedema more effectively than controlled compression therapy alone. Plastic Reconstruct Surg, 1998; 102:1058–1067.

Vignes

, Porcher

, Champagne

, Dupuy

. Predictive factors of response to intensive decongestive physiotherapy in upper limb lymphedema after breast cancer treatment: A cohort study. Breast Cancer Res Treat, 2006; 98:1–6.

Cornish

, Chapman

, Hirst

, Mirolo

, Bunce

, Ward

, Thomas

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

10.

Bar

, Cheville

, Solin

, Dutta

, Both

, Harris

. Time course of mild arm lymphedema after breast conservation treatment for early-stage breast cancer. Intl J Radiat Oncol Biol Phys, 2010; 76:85–90.

11.

Piller

, Carati

. The diagnosis and treatment of peripheral lymphedema. Lymphology, 2009; 42:146–147.

12.

International Society of Lymphology. The diagnosis and treatment of peripheral lymphedema. 2009 Concensus Document of the International Society of Lymphology. Lymphology, 2009; 42:51–60.

13.

Gjorup

, Zerahn

, Hendel

. Assessment of volume measurement of breast cancer-related lymphedema by three methods: Circumference measurement, water displacement, and dual energy x-ray absorptiometry. Lymphat Res Biol, 2010; 8:111–119.

14.

Czerniec

, Ward

, Refshauge

, Beith

, Lee

, York

, Kilbreath

. Assessment of breast cancer-related arm lymphedema—Comparison of physical measurement methods and self-report. Cancer Invest, 2010; 28:54–62

15.

Paskett

, Dean

, Oliveri

, Harrop

. Cancer-related lymphedema risk factors, diagnosis, treatment, and impact: A review. Am Soc Clin Oncol, 2012; 30:3726–3733.

16.

Shah

, Vicini

. Breast cancer-related arm lymphedema: Incidence rates, diagnostic techniques, optimal management and risk reduction strategies. Intl J Radiat Oncol Biol Phys, 2011; 81:907–914.

17.

Armer

, Stewart

. A comparison of four diagnostic criteria for lymphedema in a post-breast cancer population. Lymphat Res Biol, 2005; 3:208–217.

18.

Brorson

, Ohlin

, Olssen

, Karlsson

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

19.

Ward

, Czerniec

, Kilbreath

. Quantitative bioimpedance spectroscopy for the assessment of lymphoedema. Breast Cancer Res Treat, 2009; 117:541–547.

20.

Taylor

, Jayasinghe

, Koelmeyer

, Ung

, Boyages

. Reliability and validity of arm volume measurements for assessment of lymphedema. Phys Ther, 2006; 86:205–214.

21.

Shepherd

, Lu

. A generalized least significant change for individuals on different DXA systems. J Clin Densitom, 2007; 10:249–258.

22.

Coin

, Giannini

, Minicuci

et al. Limb fat-free mass and fat mass reference values by dual-energy X-ray absorptiometry (DEXA) in a 20 to 80 year-old Italian population. Clin Nutrit, 2012; 31:506–511.

23.

Guglielmi1

, Damilakis

, Solomou

, Bazzocchi

. A quality assurance of imaging techniques used in the clinical management of osteoporosis. Radiolog Med, 2012; 117:1347–1354.

24.

Toombs

, Ducher

, Shepherd

, De Souza

. The impact of recent technological advances on the trueness and precision of DXA to assess body composition. Obesity, 2012; 20:30–39.

25.

Baim

, Wilson

, Lewiecki

et al. Precision assessment and radiation safety for dual-energy X-ray absorptiometry: Position Paper of the International Society for Clinical Densitometry. J Clin Densit, 2005; 8:371–378.

26.

Maher

, Refshauge

, Ward

, Paterson

, Kilbreath

. Change in extracellular fluid and arm volumes as a consequence of a single session of lymphatic massage followed by rest with or without compression. Support Care Cancer, 2012; 20:3079–3086.

27.

Cornish

, Chapman

, Hirst

, Mirolo

, Bunce

, Ward

, Thomas

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