Does subcutaneous adipose tissue behave as an (anti-)thixotropic material?

ARTICLE IN PRESS

Journal of Biomechanics ] (]]]]) ]]]–]]]

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/jbiomech

Journal of Biomechanics

0021-92

doi:10.1

n Corr

Tech Ca

Tel.: +3

E-m

(M. Gee

PleasBiom

www.JBiomech.com

Does subcutaneous adipose tissue behave as an (anti-)thixotropic material?

Marion Geerligs a,b,n, Gerrit W.M. Peters c, Paul A.J. Ackermans a, Cees W.J. Oomens b, Frank P.T. Baaijens b

a Care & Health Applications, Philips Research, Eindhoven, The Netherlandsb Department of Biomedical Engineering, Technische Universiteit Eindhoven, Eindhoven, The Netherlandsc Department of Mechanical Engineering, Technische Universiteit Eindhoven, Eindhoven, The Netherlands

a r t i c l e i n f o

Article history:

Accepted 30 November 2009Although subcutaneous adipose tissue undergoes large deformations on a daily basis, there is no

adequate mechanical model to describe the transfer of mechanical load from the skin throughout the

Keywords:

Hypodermis

Mechanical properties

Rheometry

Non-linear

Long term behavior

Thixotropy

90/$ - see front matter & 2009 Elsevier Ltd. A

016/j.jbiomech.2009.11.037

esponding author at: Care & Health Applica

mpus 34, Postbox 7.1, 5656 AE Eindhoven, T

1 40 274 7977; fax: +31 40 274 6321.

ail addresses: [email protected], m

rligs).

e cite this article as: Geerligs, M., etechanics (2010), doi:10.1016/j.jbiom

a b s t r a c t

tissue to deeper layers. In order to develop such a non-linear model, a set of experimental data is

required. Accordingly, this study examines the long term behavior of adipose tissue under small strain

and its response to various large strain profiles. The results show that the shear modulus dramatically

increases to about an order of magnitude after a loading period between 250 and 1250 s, but returns to

its initial value within 3 h of recovery from loading. In addition, it was observed that the stress–strain

responses for various large strain history sequences are reproducible up to a strain of 0.15. For

increasing strains, the stress decreases for subsequent loading cycles and, above 0.3 strain, tissue

structure changes such that the stress becomes independent of the applied strain. From the results, it

can be concluded that adipose tissue likely behaves as an (anti-) thixotropic material and that a

Mooney–Rivlin model might be appropriate to simulate behavior at physiologically relevant high

strains. However, before the model is developed more fully, further experimental research is needed to

ratify that the material is (anti-)thixotropic.

& 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Mechanical load transfer from a skin contact area to deepertissues involves several tissue layers. On most body sites, thesubcutaneous adipose tissue considerably contributes to this loadtransfer. However, when numerical models are used to predictthe stress response due to external loading, the focus is either onthe skin–device contact or on the deeper tissue layers while thesubcutaneous fat layer is often ignored. This omission might berelated to the lack of defined parameters that describe themechanical behavior of adipose tissues. This is particularlysurprising given the critical roles for adipose tissues in themedical and cosmetic fields, involving, for example, implantabledrugs delivery, skin adhesive removal, deep tissue injury andneedle insertion procedures.

Recently, our previous work on the linear behavior ofsubcutaneous adipose tissue has shown that the linear strainregime is valid for very small strains only, i.e. 0.001 (Geerligset al., 2008). In most applications, however, much higherdeformations occur in the adipose tissue for prolonged periods.Indeed, for wheelchair or bedridden patients, for example, this

ll rights reserved.

tions, Philips Research, High

he Netherlands.

[email protected]

al., Does subcutaneous adech.2009.11.037

might lead to the development of deep tissue injury under bonyprominences within a time frame of minutes to hours, duringwhich stress relaxation in the compressed tissue might occur(Gefen and Haberman, 2007). Numerical models based onexperimental data are of indispensable value to predict the onsetand progression of such mechanically induced damage.

Currently, there is a paucity of papers on the mechanicalproperties of subcutaneous adipose tissue found beneath hairyskin. Viscoelastic properties of single human adipocytes have beenrecently characterized using AFM, resulting in a relaxed modulusand relaxation time for either load or deformation (Darling et al.,2008). A few related in vitro studies on tissue behavior exist.Of these, rheological measurements demonstrated a decrease inviscosity with increasing shear rate (Patel et al., 2005). In addition,the authors suggested that adipose tissue loses firmness withincreasing strain and frequency, a state that is not recoverable. In aseparate study, ovine subcutaneous tissue was subjected to ramp-and-hold cycles during confined compression tests at variousramp rates (Patel et al., 2005; Gefen and Haberman, 2007). Theresults were given in the form of a transient aggregate modulusand short-term elastic moduli. They also found a strong deforma-tion rate dependency. Short-term moduli were in the order of20 kPa. In an alternative in vivo approach, a suction device yieldedexperimental parameters, which, when combined with numericalmodeling, led to a first estimation of non-linear material param-eters for human skin (Hendriks et al., 2003). To our knowledge,there are no in vivo studies considering subcutaneous adipose

ipose tissue behave as an (anti-)thixotropic material? Journal of

ARTICLE IN PRESS

M. Geerligs et al. / Journal of Biomechanics ] (]]]]) ]]]–]]]2

tissue as a single layer. By contrast, some in vivo studies haveexamined the mechanical properties for a compliant systemconsisting of skin and subcutaneous adipose tissue (Linder-Ganzet al., 2007; Then et al., 2007).

The work mentioned above describes a range of loadingconditions, often combining techniques involving indentation,confined compression, stress relaxation and constant shearresponses. Clearly, this makes comparison of data from thestudies problematic in nature. For a general constitutive modelfor adipose tissue a more systematic approach is required.

The material structure of subcutaneous adipose tissue does notrelate conveniently to those of other biological tissues. Its maincomponent is the white adipocyte. The remaining components arewater (5–30% weight) and protein (2–3% weight). The whiteadipocytes are filled with a large fat droplet imposing forces onboth the nucleus and the small cytoplasmic volume at the cellperiphery. The composition of the white adipocytes depends onthe specific function and body site. As an example, differencesthroughout the human body are known for the proportionsof saturated fatty acids, monosaturated vs. polysaturated fatand the lipolysis rate (Avram et al., 2005). White adipocytes arecollected in a surrounding fiber network. The adipose tissueis well vascularized throughout with each adipocyte in contactwith at least one capillary. Hence, adipose tissue is susceptibleto ischemia and hypoxia, which influence its mechanicalresponse.

Our previous work on the small strain behavior of adipose tissuehas shown that reproducible results are obtained in an in-vitro

set-up using a rheometer with parallel plate geometry and thatthe behavior can be described with a power-law model (Geerligset al., 2008). However, sometimes tissue samples were found to bemuch stiffer than the mean value and early work at higher strainshas suggested that (reversible) structural changes start to play arole. In addition, earlier large strain studies formed the incentivefor a more systematic approach at higher strains to elucidate thephenomena that have already been described. Therefore, thepresent study aims to provide systematic data for long-termsmall strain behavior as well as the effect of strain history, withthe purpose of contributing to the development of a constitutivemodel.

Accordingly, the work is divided in two parts. The first partcontains long term oscillatory tests at small strains to investigatetemporal effects of the adipose tissue samples. Subsequently,strain-dependency tests, comprising constant shear, stress relaxa-tion and constant strain rate, are applied. From these tests, non-linear parameters can be obtained that are useful for constitutivemodeling. Such an experimental approach is designed to gaininsight on the mechanical response of adipose tissue under shearwhere the effect of strain history, strain level and duration istaken into account.

2. Materials and methods

2.1. Sample preparation

In porcine species, the subcutaneous fat layer on the back is divided in an

outer, middle and inner layer. The porcine middle layer was selected for use, as it

is considered to be the most comparable with the deep subcutaneous layer in the

abdominal region of humans (Klein et al., 2007). The tissue was obtained from a

local slaughterhouse, where they were cut into transverse slices of approximately

1.5 mm thick. In our laboratories, circular samples were obtained from the slices

with an 8 mm diameter cork borer. The samples were stored in a phosphate

buffered saline solution (PBS) in ice-cooled boxes and tested within 48 h of

collection. If measurements were repeated after a certain period of recovery, each

sample was stored in PBS between measurements. All pigs were Landrace, having

a dressed carcass weight of approximately 83 kg, and were 14–18 weeks old at

necropsy.

Please cite this article as: Geerligs, M., et al., Does subcutaneous adBiomechanics (2010), doi:10.1016/j.jbiomech.2009.11.037

2.2. Rheological methods

All experiments were performed on a rotational rheometer (ARES, Rheometric

Scientific, USA) with parallel plate geometry in combination with a Peltier

Environmental control unit and a fluid bath. Plates were sand-blasted to prevent

slippage. The upper plate was lowered to compress the sample until the sample

experienced an axial force of 1 g. All loading protocols, which were based on

previous experiments on soft biological tissues (van Dam et al., 2008; Hrapko

et al., 2006), are summarized in Fig. 1.

Long-term dynamic behavior within the linear viscoelastic regime was studied

with time sweep tests (Fig. 1a). Tests were performed at a frequency of 10 rad/s

with a strain amplitude of 0.001 at body temperature (37 1C), lasting at least 45 min.

The chosen strain amplitude was previously determined to be the maximum strain

within the linear viscoelastic regime (Geerligs et al., 2008). Time sweeps were

repeated after various time periods of recovery, namely 0, 0.5, 1 and 3 h.

Shear experiments in the non-linear regime were preceded by two successive

frequency sweeps with a frequency of 1–100 rad/s and a strain amplitude of 0.001.

This procedure was adopted to minimize the effects of pre-conditioning (Geerligs

et al., 2008). Subsequently, the sample was tested in either a series of constant

shear rate experiments, constant shear experiments or stress relaxation experi-

ments (Fig. 1b–e). The measurement protocols were based on previous experi-

ments on soft biological tissues.

Constant shear rate experiments with various strain amplitudes (Fig. 1b and c)

were designed to investigate any potential damaging effect in the mechanical

behavior due to the previous strain history on immediate mechanical response. The

first series of sequences were loading–unloading tests conducted with a constant

shear rate of 1 s�1 and strains incrementally increasing from 0.01 up to 0.5 (Fig. 1b).

The sample was left to recover at zero strain for at least 10 times the loading time

after each loading–unloading cycle. In total, 20 cycles were applied. In another series

of sequences with the same constant shear rate, strains were applied in decreasing

order (Fig. 1c). Again the sample was left to recover at zero strain for at least 10 times

the loading time after each loading–unloading cycle. In order to investigate possible

reversible changes, this sequence was repeated after 0, 1 and 3 h of rest.

The next set of experiments was designed to apply constant shear at

increasing shear rate (Fig. 1d). Loading–unloading cycles were conducted with

constant shear rate increasing from 0.01 to 1 s�1 per cycle with maximum strain

amplitude of 0.15. Between two cycles, the sample was again left to recover for at

least 10 times the loading time.

Finally, stress relaxation experiments were composed of a series of ramp-and-

hold tests at different strain levels (Fig. 1e). During the loading and unloading

phase, a constant strain rate of 1 s�1 was imposed. The maximum strain was held

for 10 s, during which relaxation of the material was recorded. The sample was left

to recover for a period of at least 100 s, during which time the tissue response was

monitored. The test was repeated for four different strain levels, namely 0.01, 0.05,

0.1 and 0.15.

An overview of the number of specimens and the number of samples from

each specimen per test is given in Table 1.

3. Results

3.1. Long term small strain behavior

An interesting qualitative trend was observed during the timesweep experiments (Fig. 2a). The samples showed a gradualincrease of both initial storage modulus and initial loss moduliover time from the start of the experiment. However, after aperiod, a rapid increase in stiffness, DG0, occurred in all samples,indicating a change in tissue structure. The moduli showed afurther slight increase until a steady state was reached. During thesteep increases the moduli increased by a range of roughly1.5–15 kPa. The rapid stiffening occurred at some time between250 and 1200 s. An overview of the stiffness increase and starttime for all 13 samples is given in Fig. 2b.

Experiments with repeated time sweeps show that the materialbehavior is reversible, although recovery takes several hours tocomplete (Fig. 3). To enable comparison between specimens, theshear moduli of each specimen were normalized to a scale r from 0to 1, e.g. from the initial modulus up to the final steady state levelof the initial test. When the second time sweep is immediatelyperformed after the first time sweep, the initial moduli remainconstant at the plateau value, see Fig. 3a. After a recovery period of1 h, the initial value for the moduli is reduced, although notreaching the level corresponding to that during the first time

ipose tissue behave as an (anti-)thixotropic material? Journal of

ARTICLE IN PRESS

Fig. 1. Schematic illustration of test sequences. (a) Time sweep tests, (b) constant shear rate experiments with increasing shear strains, (c) constant shear rate experiments

with decreasing shear strains, (d) constant shear experiments with increasing shear rate and (e) stress relaxation experiments.

Table 1Overview of number of samples used for the experiments: the number of

specimens per test (the number of samples from each specimen).

Test # specimens (# samples per specimen)

Time sweep 4 (1,6,4,2)

Constant shear rate

Increasing shear 1 (3)

Decreasing shear 1 (2)

Constant shear 2 (3,3)

Stress relaxation 2 (3,3)

M. Geerligs et al. / Journal of Biomechanics ] (]]]]) ]]]–]]] 3

Please cite this article as: Geerligs, M., et al., Does subcutaneous adBiomechanics (2010), doi:10.1016/j.jbiomech.2009.11.037

sweep. After 3 h the material appeared to be totally recovered anda qualitatively comparable curve could be obtained. A third test onthe same sample after a further 3 h of recovery (trest=6 h in Fig. 3b)demonstrated a qualitatively similar curve.

3.2. Large strain experiments

In the constant shear rate experiment with increasing strains(Fig. 4a), three phases can be distinguished as delineated by strainvalues of 0.15 and 0.30 in Fig. 4b. If the stress–strain curve

ipose tissue behave as an (anti-)thixotropic material? Journal of

ARTICLE IN PRESS

101 102 103 101 102 103103

104

105

103

104

105

G’ [

Pa]

begin time transition [s]

ΔG’ [

Pa]

subject 1subject 2subject 3subject 4

time [s]

ΔG’

Fig. 2. (a) Typical result of a time sweep. The arrow indicates the measured increase in the storage modulus G0 during quick stiffening phase (DG0). (b) DG0 against the start

time of stiffening for samples from all specimens.

103

104

105

G’ [

Pa]

t0trest = 3 hrtrest = 6 hr

101 102 1030

10 hr

0.5 hr

1 hr

3 hr

r [−]

time [s]101 102 103

time [s]

Fig. 3. Repetition of time sweeps. (a) The shear moduli are scaled from 0 to 1, from the start value of the initial test on the specific sample up to the stationary state at the

higher plateau. The initial response from one sample is shown here by the thick line; the other lines represent the response after various periods of rest time for the same

sample. (b) A sample is loaded again after 3 and 6 h of rest to demonstrate the reversible behavior.

0 0.1 0.2 0.3 0.4 0.5

−200

0

200

400

600

shear strain � [−]

shea

r stre

ss τ

[Pa]

I

0 0.5 10

0.1

0.2

0.3

0.4

0.5

shea

r stra

in �

[−]

time [s]0 0.05 0.1

−100

0

100

200

300

400

500

shear strain � [−]

shea

r stre

ss τ

[Pa]

Fig. 4. Average results from constant shear rate experiment with increasing strain amplitude. (a) Applied shear strain with reproducible strain rate; (b) the three different

phases of the stress–strain response; (c) stress–strain response up to 0.1% strain.

M. Geerligs et al. / Journal of Biomechanics ] (]]]]) ]]]–]]]4

(Fig. 4b) is enlarged to highlight the first phase, it is evident thatthe responses at strains up to 0.15, within reasonable limits,overlap (Fig. 4c) and can be considered to be reproducible.For strains above 0.15, however, the loading curves change.

Please cite this article as: Geerligs, M., et al., Does subcutaneous adBiomechanics (2010), doi:10.1016/j.jbiomech.2009.11.037

For increasing strain, the stress decreases for subsequent loadingcycles, indicating strain induced changes in the tissue. By contrast,above 0.3 strain, the curves appear to overlap for repeated loadcycles, suggesting that tissue structure does not change further.

ipose tissue behave as an (anti-)thixotropic material? Journal of

ARTICLE IN PRESS

0 0.5 10

0.1

0.2

0.3

0.4

0.5

shea

r stra

in, �

[−]

time [s]0 0.05 0.1 0.15 0.2 0.25 0.3

−100

−50

0

50

100

150

200

shea

r stre

ss, �

[Pa]

0 hr

1 hr

3 hr

shear strain, � [−]

Fig. 5. Average results from constant shear rate experiment with decreasing strain amplitude. (a) Applied shear strain with reproducible strain rate. (b) Stress–strain

curves from constant shear rate experiments with decreasing strain. The applied sequences have been repeated after various rest periods (dotted lines).

0 0.05 0.1 0.15

−100

0

100

200

300

400

500

shea

r stre

ss τ

[Pa]

0.01 s−1

0.1 s−1

1 s−1

shear strain � [−]

Fig. 6. Constant shear experiments with increasing strain rate.

M. Geerligs et al. / Journal of Biomechanics ] (]]]]) ]]]–]]] 5

Although the stress response greatly differs for the three phasesfor the large strain range, the stress response within the linearstrain region did not change.

The results of the constant shear rate experiments withdecreasing strain are depicted in Fig. 5. Notice that the tissuestructure immediately changed in the first cycle, and that thesubsequent loading cycles followed the first curve. In addition,despite applying strains of approximately 0.3, the specimens wereable to recover after a sufficient recovery period.

Constant shear rate experiments with increasing strain ratewere applied up to a maximum strain of 0.15. From the results itcan be observed that the stress as a function of strain is strain ratedependent and that the response stiffens with increasing strainrate for both linear and non-linear ranges (Fig. 6).

Results of the stress relaxation experiments are illustratedin Fig. 7. The results show practically overlapping curves for theloading phase in the linear strain regime (Fig. 7b). The stress

Please cite this article as: Geerligs, M., et al., Does subcutaneous adBiomechanics (2010), doi:10.1016/j.jbiomech.2009.11.037

response in the non-linear strain region followed a nearly iden-tical curve for each sample (Fig. 7c). During stress relaxation,the relaxation modulus did not reach yet a plateau value withinthe relaxation time allowed (Fig. 7d). The averaged relaxationmodulus decreases as a function of applied strain, where thedifference becomes smaller for larger strains.

4. Discussion

For this study, both long term behavior at small strains andstrain history effects at large strains were investigated. Samplesfrom porcine subcutaneous adipose tissue demonstrated note-worthy behavior for both types of loading. The long term behaviorobtained at small strains is qualitatively reproducible. However,in quantitative terms, both the time of onset and the amount ofincrease in moduli values varied considerably (Fig. 2b). The causefor these variations is not yet understood. Nevertheless, theobserved sudden stiffening of the material up to a decade iscrucial for understanding and measuring the material behavior ofadipose tissue. The rapid increase in tissue stiffness impliesstructural changes, which are reversible, and might influencemechanical testing over longer time periods.

Responses in the large strain regime were examined initially byperforming constant shear rate experiments (Fig. 1b). The stress–strain response changed for increasing strains and can be dividedinto three phases (Fig. 4b). Material behavior changed dramati-cally. Additional experiments were therefore performed to ratifythe tissue structure changes due to mechanical loading, as well asto investigate tissue recovery. These experiments with decreasingshear confirmed that the stress–strain response is dependent onthe strain history. The applied large strains here are in accordancewith physiologically relevant strains, for example equivalent tothat estimated during sitting (Linder-Ganz et al., 2007).

From the constant shear rate experiments it can be concludedthat up to 0.15 strain, the adipose tissue might behavemechanically similar to other biological tissues such as braintissue and thrombus (van Dam et al., 2008; Hrapko et al., 2006).Because tissue structure changes might occur above 0.15 strain,

ipose tissue behave as an (anti-)thixotropic material? Journal of

ARTICLE IN PRESS

0 5 10−100

0

100

200

300

time [s]

shea

r stre

ss, �

[Pa]

shea

r stre

ss, �

[Pa]

γ = 0.01

γ = 0.05

γ = 0.10

γ = 0.15

0 0.05 0. 1 0.15−100

0

100

200

300

0 0.05 0. 1 0.15−100

0

100

200

300

shear strain, � [−]

shea

r stre

ss, �

[Pa]

10−1 100 101102

103

104

Gt [

Pa]

γ = 0.01

γ = 0.05

γ = 0.10

γ = 0.15

shear strain, � [−]

time [s]

Fig. 7. Results of stress relaxation experiments in shear (test sequence C). (a) Stress vs. time for one sample, (b) stress–strain response for one sample, (c) peak stress

variations (n=6) and (d) average relaxation modulus vs. time.

M. Geerligs et al. / Journal of Biomechanics ] (]]]]) ]]]–]]]6

the subsequent large strain experiments were performed upto this limit. The constant shear experiments and stress relaxa-tion tests indicate both reliability and reproducibility of thetest method and show similar trends as those reported forsamples from brain and thrombus tissues. These findings there-fore support the appropriateness of a Mooney–Rivlin like modelfor the simulation of the first phase of large strains.

Structural changes due to mechanical loading are an indicationof thixotropic behavior. Thixotropic behavior is defined as a time-dependent decrease of viscosity or modulus induced by deforma-tion that is a reversible effect when the deformation is removed(Mewis and Wagner, 2009). When the deformation causes areversible, time-dependent increase, it is called antithixotropy.(Anti-)thixotropic materials may or may not be viscoelastic innature. Both the long term behavior at small strains and theconstant shear rate experiments indicate reversible structuralchanges. However, the small strain results indicate an anti-thixotropic behavior, while the large strain results show athixotropic behavior that is observed at large strains only. Thestress relaxation response evidently indicates viscoelastic beha-vior. In the human body, blood and synovial fluid are known tobehave thixotropically (Mewis and Wagner, 2009; O’Neill andStachowiak, 1996). For adipose tissues, it would be interesting tovisualize using a confocal microscope to see whether adipocytesor/and the surrounding collagen network rearrange with mechan-ical loading. In addition, to examine the mechanical behavior for

Please cite this article as: Geerligs, M., et al., Does subcutaneous adBiomechanics (2010), doi:10.1016/j.jbiomech.2009.11.037

strains above 0.15 specific test methods are needed, as summar-ized in a recent overview (Mewis and Wagner, 2009). Whenestablishing such experiments, the large strain behavior ofadipose tissues should be studied preferably before stiffeningoccurs at small strains to be independent of time effects.

The outcomes of our large strain studies were not influenced bytime effects. From the large deformation studies, the experimentwith an increasing strain up to 50% represented the most prolongedlasting approximately 2300 s, including the preceding frequencysweeps. The loading–unloading cycle was maintained at a maximumstrain for only 1 s, which amounted to only 20 s in total. The durationof the other experiment with increasing strains was less than 500 s.The increasing shear rate experiments and stress relaxation experi-ments lasted approximately 900 and 750 s with short term loading–unloading cycle as well. So the long-term time effects did notinfluence the outcome of the strain-dependency studies.

The observed reversible behavior is in contradiction with aprevious study (Patel et al., 2005). These authors argue that evenat small deformations human adipose tissue is not able to recoverduring creep tests. Since the linear strain regime is applicable onlyto very small strains, it might be that those measurements areperformed outside this region or that the recovery time wasinsufficient.

The described phenomena may have major consequences forthe interpretation of results of biomechanical studies. A field ofinterest of the authors is the development of pressure ulcers,

ipose tissue behave as an (anti-)thixotropic material? Journal of

ARTICLE IN PRESS

M. Geerligs et al. / Journal of Biomechanics ] (]]]]) ]]]–]]] 7

tissue degeneration after prolonged loading, usually occurring inbedridden or wheelchair bound patients. Recent studies haveshown that these ulcers can start at the skin, but also in deepertissue layers close to bony prominences (Linder-Ganz and Gefen,2009; Stekelenburg et al., 2008). This pressure induced ‘‘deeptissue injury’’ is a major issue for wheelchair bound paraplegicpatients because they are insensate to pressure-induced effectsand injury is very difficult to diagnose in the absence of visibledamage at the skin surface. In studies on etiology and develop-ment of methods for prevention, biomechanical modeling is avaluable tool. The fat layer plays a very important role in theseanalyses and the stiffness changes described in the current paperwill have a major impact on the stress and strain distributionswithin the different tissue layers overlying the bony prominences.This highlights the need for further research on this subject and toderive a theoretical model for the description of fat behavior.

In conclusion, the time sweeps tests and the large strainexperiments demonstrate that time effects and strain effects resultin different material behaviors. This indicates (anti-)thixotropicmaterial behavior meaning that a constitutive model should containparameters to describe the build-up and breakdown of materialstructure. When only large strains up to 0.15 are considered, aMooney–Rivlin model should be able to capture the experimentaldata. The application of the Mooney–Rivlin model would demandextra parameters to include the effect of prolonged mechanicalloading as well as the physiologically relevant high strains.Additionally, a power law model describing the linear viscoelasticbehavior has been introduced in our previous work. This modelwould also be suitable for implementing a build-up and breakdownstructure properties. We believe, however, it is better to set up moreexperiments to fully understand the material behavior beforecontinuing the building of a constitutive model.

This paper shows the high complexity of material behavior andparticularly demonstrates that more work is needed on this topic.The described effects should be taken into account when settingup new experiments. The follow-up experiments should clarifythe effects of time and strain and the reversibility of the material.

5. Conflict of interest

The authors and the author’s institutions do not have a conflictof interest due to any financial or other relationship with otherpeople or organizations.

Please cite this article as: Geerligs, M., et al., Does subcutaneous adBiomechanics (2010), doi:10.1016/j.jbiomech.2009.11.037

Acknowledgements

We would like to thank Prof. Dan Bader for his valuablecontribution to our discussions during preparation of this article.

References

Avram, A.S., Avram, M.M., James, W.D., 2005. Subcutaneous fat in normal anddiseased states: 2. Anatomy and physiology of white and brown adipose tissue.Journal of the American Academy of Dermatology 53, 671–683.

Darling, E.M., Topel, M., Zauscher, S., Vail, T.P., Guilak, F., 2008. Viscoelasticproperties of human mesenchymally-derived stem cells and primary osteo-blasts, chondrocytes, and adipocytes. Journal of Biomechanics 41, 454–464.

Geerligs, M., Peters, G.W., Ackermans, P.A., Oomens, C.W., Baaijens, F.P., 2008.Linear viscoelastic behavior of subcutaneous adipose tissue. Biorheology 45,677–688.

Gefen, A., Haberman, E., 2007. Viscoelastic properties of ovine adipose tissuecovering the gluteus muscles. Journal of Biomechanical Engineering—Transac-tions of the ASME 129, 924–930.

Hendriks, F.M., Brokken, D., van Eemeren, J.T.W.M., Oomens, C.W.J., Baaijens, F.P.T.,Horsten, J.B.A.M., 2003. A numerical-experimental method to characterize thenon-linear mechanical behaviour of human skin. Skin Research and Technol-ogy 9, 274–283.

Hrapko, M., van Dommelen, J.A., Peters, G.W., Wismans, J.S., 2006. The mechanicalbehaviour of brain tissue: large strain response and constitutive modelling.Biorheology 43, 623–636.

Klein, J., Permana, P.A., Owecki, M., Chaldakov, G.N., Bohm, M., Hausman, G.,Lapiere, C.M., Atanassova, P., Sowinski, J., Fasshauer, M., Hausman, D.B.,Maquoi, E., Tonchev, A.B., Peneva, V.N., Vlachanov, K.P., Fiore, M., Aloe, L.,Slominski, A., Reardon, C.L., Ryan, T.J., Pond, C.M., 2007. What are subcuta-neous adipocytes really good for y? Experimental Dermatology 16, 45–70.

Linder-Ganz, E., Gefen, A., 2009. Stress analyses coupled with damage laws todetermine biomechanical risk factors for deep tissue injury during sitting.Journal of Biomechanical Engineering 131, 011003.

Linder-Ganz, E., Shabshin, N., Itzchak, Y., Gefen, A., 2007. Assessment ofmechanical conditions in sub-dermal tissues during sitting: a combinedexperimental-MRI and finite element approach. Journal of Biomechanics 40,1443–1454.

Mewis, J., Wagner, N.J., 2009. Thixotropy. Advances in Colloid and Interface Science147–148, 214–227.

O’Neill, P.L., Stachowiak, G.W., 1996. The inverse thixotropic behaviour of synovialfluid and its relation to the lubrication of synovial joints. Journal ofOrthopaedic Rheumatology 9, 222–228.

Patel, P.N., Smith, C.K., Patrick, C.W., 2005. Rheological and recovery properties ofpoly(ethylene glycol) diacrylate hydrogels and human adipose tissue. Journalof Biomedical Materials Research Part A 73A, 313–319.

Stekelenburg, A., Gawlitta, D., Bader, D.L., Oomens, C.W., 2008. Deep tissue injury:how deep is our understanding? Archives of Physical Medical Rehabilitation89, 1410–1413.

Then, C., Menger, J., Benderoth, G., Alizadeh, M., Vogl, T.J., Hubner, F., Silber, G.,2007. A method for a mechanical characterisation of human gluteal tissue.Technology Health Care 15, 385–398.

van Dam, E.A., Dams, S.D., Peters, G.W., Rutten, M.C., Schurink, G.W., Buth, J., van deVosse, F.N., 2008. Non-linear viscoelastic behavior of abdominal aorticaneurysm thrombus. Biomechanics and Modeling in Mechanobiology 7,127–137.

ipose tissue behave as an (anti-)thixotropic material? Journal of