Fracture toughness of human amniotic membranes

Abstract

Amnion is a membrane that surrounds and structurally protects the developing fetus during pregnancy. The rupture of amniotic membranes prior to both normal and preterm deliveries involves stretch forces acting on a biochemically triggered weak zone of the membranes. Fracture toughness is an important mechanical property describing how the membranes containing a defect resist fracture, but this property has never been investigated in amniotic membranes. In this work, the fracture toughness of many samples cut from four pieces of amniotic membrane from different mothers was examined by uniaxial and pure shear (mode I) fracture tests. The measurement was checked for dependence on the sample geometry and notch length. Results from the uniaxial tensile test show J-shaped stress–strain curves and confirm that the amniotic membrane is a nonlinear material. The measured fracture toughness of four amniotic membranes ranged from 0.96 ± 0.11 to 1.83 ± 0.18 kJ m⁻². Despite considering the effect of the presence of the defect on mechanical property measurement, similar fracture behaviour was observed for pre-notched and unnotched specimens, indicating that the membranes were extremely tolerant to defects. This defect-tolerant characteristic provides insight into the understanding of fetal membrane rupture.

1. Introduction

The fetal membrane, consisting of two layers, the chorion and the amnion, is a soft tissue surrounding a fetus during pregnancy [1]. The amnion retains amniotic fluid and is mainly composed of collagen fibres, providing most of the structural support surrounding the fetus [2]. The rupture of fetal membranes occurs in both normal and preterm deliveries. Premature rupture of membranes (PROM) is when the membrane ruptures before labour begins. PROM is an important process in pregnancy because it induces labour by allowing the release of amniotic fluid. The rupture of membranes prior to full gestation is known as preterm premature rupture of membranes (PPROM), and it occurs before 37 weeks gestation. Both of the conditions, either PROM in normal delivery or PPROM in abnormal conditions, require an understanding of the fracture of fetal membranes.

The detailed mechanism and the causes of idiopathic rupture of the membrane are not well understood. Existing studies investigated repeated stretching from uterine contractions and the viscoelastic deformation of membrane rupture [3–5]. In addition to the mechanical loading, membrane rupture is a consequence of a weakening process involving separation of fetal membrane layers [6] and a formation of a weak zone that has decreased thickness and weaker mechanical properties as a result of collagen remodelling and cellular apoptosis [7–9].

Amnion is often considered as a nonlinear and time-dependent material and has been characterized by uniaxial and biaxial tensile testing [4,10–12], relaxation and cyclic testing [13], inflation testing [14–16] and puncture tests [2,17,18]. Existing studies quantify membrane rupture with material parameters including strength and work of rupture. The membrane rupture was found to relate to its thickness and membrane layer integrity [14,19].

While the weakening process involves the formation of a zone with altered morphology, it is crucial to consider the existence of defects in the study of membrane rupture. One key step towards understanding the rupture of fetal membranes is based on fracture mechanics approach that studies how the existence of a flaw induces material failure. A notch is introduced to the specimens prior to fracture testing. The measurement of fracture toughness quantifies the energy release per unit of new crack area propagated from the notch [20].

This work aims to examine the fracture toughness of amniotic membranes and obtain a better understanding of how pre-existing flaws on the membrane influence fracture. Uniaxial tensile and fracture tests were performed on four pieces of amniotic membranes to characterize their fracture toughness and compare the deformation and fracture behaviour of unnotched and pre-notched specimens. The measurement of fracture toughness was validated by testing a range of samples with different notch lengths and sample dimensions. This study provides quantitative measurement of the fracture toughness of human amniotic membranes and a basic understanding of fracture resistance of the amniotic membrane.

2. Material and methods

2.1. Specimen preparation

Four full-term chorioamniotic membranes were obtained via the Pregnancy Outcomes Prediction Study (POPS) at the Rosie Hospital, Cambridge. All patients gave informed consent for the donation of the otherwise discarded membranes. The membranes were collected from the patients who were unlaboured and kept in conical tubes in a freezer at −40°C. The frozen membranes were thawed by submerging the conical tubes containing the membrane in a beaker containing warm tap water. Phosphate-buffered saline (PBS) solution was added to the tubes to hasten the thawing process.

The amnion layer was then separated gently from the chorion and stored in PBS solution for later cutting and testing. First, a template consisting of the exact sample geometries was drawn in Adobe Illustrator CS5 and printed on a normal paper (figure 1). The template consists of sample size with a gripping region at both top and bottom edges. The template was hydrated using PBS before putting on the amnion membrane (AM) to avoid any traction force between template and the membrane. Given that the AM is translucent, the geometry shown in the template was visible even when the AM was placed on top. Strips of sandpaper were then stuck at the gripping region on the top and bottom surfaces of the samples by applying super glue. The AM was then cut according to the template using scissors. The notch in the sample, also shown in the template, was cut with a sharp scalpel.

Figure 1.

Procedure to prepare samples of a human AM for mechanical testing: (a) printed sample template, (b) AM placed on top of the template, (c) super glue applied to a sandpaper strip, (d) sandpaper carefully attached to the AM at the grip region, (e) sandpaper attached at both upper and lower surfaces of grip region, (f) AM cut according to the template, (g) notch cut in the sample using a scalpel and (h) cut samples in a Petri dish containing PBS solution ready for testing. (Online version in colour.)

2.2. Mechanical testing

Mechanical tests were performed on an Instron 5544 universal testing frame (Canton, MA, USA) with a 500 N load cell. There were 57 samples cut from four pieces of AM (AM1, AM2, AM3, AM4) obtained from different mothers. Three AM (AM1, AM2, AM3) for planar tests (PT) and planar fracture (PF) tests were prepared by cutting 10–13 rectangular strips from each AM with dimension of 36 × 4 mm (table 1). A notch with a length of 13 or 18 mm was introduced to the fracture samples with a sharp scalpel prior to the test. The samples were clamped along the long edges to measure the properties in the planar configuration. Eleven strips with a dimension of 5 × 11 mm were cut from the fourth membrane (AM4) for tensile tests (TT) and tensile fracture (TF) tests. A notch of 1.5 mm was introduced in the TF samples prior to the test. In contrast with planar testing, the tensile samples were clamped along the specimen's short edges. Twelve planar samples with a smaller width (23 × 4 mm) compared to previous planar samples were also cut from AM4 to check measurement consistency with respect to sample dimensions.

Table 1.

The sample sizes for tensile (TT), tensile fracture (TF), planar (PT) and planar fracture (PF) tests on four human amniotic membranes (AM1, AM2, AM3 and AM4) from different mothers.

name	test type	width, w (mm)	length, l (mm)	notch length, a (mm)	sample number, n
AM1_PT	planar test	36	4	0	6
AM1_PF1	planar fracture test	36	4	13	3
AM1_PF2	planar fracture test	36	4	18	4
AM2_PT	planar test	36	4	0	5
AM2_PF1	planar fracture test	36	4	13	3
AM2_PF2	planar fracture test	36	4	18	3
AM3_PT	planar test	36	4	0	5
AM3_PF1	planar fracture test	36	4	13	5
AM4_PT	planar test	23	4	0	5
AM4_PF1	planar fracture test	23	4	8	7
AM4_TT	tensile test	5	11	0	6
AM4_TF	tensile fracture test	5	11	1.5	5

All samples were deformed at a constant rate of 0.05 mm s⁻¹. The samples were slightly slack at the beginning of the test to prevent the application of pre-stress in the samples before the tests. The samples were also sprayed with PBS prior to testing to avoid dehydration. The first displacement was when the stress was greater than σ₀ = 0.028 MPa: PT with 36 × 4 mm size, for example, have P₀ = 0.05 N, 23 × 4 mm size have P₀ = 0.03 N and tension tests with 5 × 11 mm size have P₀ = 0.007 N as the reference for zero displacement.

2.3. Data analysis

When the specimens were pulled apart by the universal tensile machine, the force F and displacement Δ were recorded. The strain $\epsilon$ was determined by dividing the displacement measured from the experiment by the initial length l₀. All stresses $\sigma$ involved in this study were determined by dividing the force by the area without notch length a. The parameters including material strength or failure stress σ_f, failure strain ɛ_f and fracture toughness G_c were examined. The failure stress σ_f and failure strain ɛ_f were taken at the points of maximum stress found in the stress–strain curves. Material strength is defined as the failure stress of an unnotched specimen.

Fracture toughness G_c is determined following classical studies of elasticity and fracture mechanics of rubbers proposed by Rivlin & Thomas [21]. The toughness G_c is determined by analysing the measurement of the unnotched and the pre-notched specimens with identical dimension in PT and PF testing (equation (2.1)).

$$G_{\mathrm{c}}=w_0{l}_0.$$ 2.1

The energy $w_0$ is the energy per unit volume absorbed of the unnotched sample (the area under stress–strain curves) in the PT test, corresponding to the failure strain ${\epsilon }_{\mathrm{f}}$ obtained for the pre-notched samples in PF test.

As a comparison of fracture criterion G_c used by Rivlin & Thomas, here we also examined a value of characteristic energy $E_{\mathrm{f}}$ involving energy density stored in each sample up to the fracture point (equation (2.2)).

$$E_{\mathrm{f}}=w_{\mathrm{f}}{l}_0.$$ 2.2

The strain energy density $w_{\mathrm{f}}$ is obtained by calculating the areas under the stress–strain curves up to the fracture points ${\epsilon }_{\mathrm{f}}$ of each test (equation (2.3)). The $\sigma$ and $\epsilon$ are stress and strain of the specimens.

$$w_{\mathrm{f}}={\int }_0^{{\epsilon }_{\mathrm{f}}}\sigma \mathrm{d}\epsilon .$$ 2.3

Both G_c and $E_{\mathrm{f}}$ were obtained by multiplying the strain energy density up to the fracture points and the initial sample length. The difference between G_c and $E_{\mathrm{f}}$ is that G_c was determined from both PT and PF testing while $E_{\mathrm{f}}$ only involved a single type of testing. The strain energy density of G_c involves the integration of the area under stress–strain curves of PT samples up to the PF fracture points. The strain energy density of $E_{\mathrm{f}}$ involves the integration of stress–strain curves up the fracture points of each sample. All areas under the curves were obtained using commercial integration tools in Origin (OriginLab, Northampton, MA).

3. Results

3.1. Mechanical properties of amniotic membranes

Figure 2 shows the stress–strain curves of all samples tested from four pieces of amniotic membrane. Both planar and tensile testing produce nonlinear J-shaped stress–strain curves, with very compliant responses at the beginning and an increase of moduli when the strain increases. The maximum stresses and strains of the stress–strain curves were identified as their fracture points of the specimens. The fracture stresses of unnotched samples for both TT and PT tests were recorded as the strength (σ_f) of AM. The slope on the linear region close to these fracture points was taken to be the elastic modulus (E) of the membranes. The areas under the stress–strain curves of PT corresponding to the fracture strain of PF tests were used to determine the fracture toughness (G_c) of AM. Table 2 shows mechanical properties for four pieces of AM. The strength in planar configuration ranged from 1.34 ± 0.55 to 2.00 ± 1.19 MPa, elastic modulus ranged from 2.40 ± 0.46 to 5.11 ± 1.55 MPa while fracture toughness varied from 0.96 ± 0.11 to 1.83 kJ m⁻². Only AM4 was tested in tensile condition providing slightly larger modulus of 7.37 ± 1.76 MPa and similar strength of 1.91 ± 0.66 MPa as compared to its properties in the planar configuration.

Figure 2.

(a) Schematic illustration of sample geometries for planar (PT), planar fracture (PF), tensile (TT) and tensile fracture (TF) tests. The planar tests have two specimen sizes with 36 × 4 mm for Geo1 and 23 × 4 mm for Geo2. All fracture specimens have a notch length a. (b–e) Stress–strain (σ–ɛ) curves of PT and PF tests for four amniotic membranes (AM1, AM2, AM3 and AM4). (f) Stress–strain curves of TT and TF tests for AM4. All circles show the fracture points of the corresponding curves. (Online version in colour.)

Table 2.

Strengths, elastic moduli and fracture toughnesses of four pieces of AM (AM1, AM2, AM3 and AM4) obtained from planar (PT, PF) and tensile (TT and TF) testing.

sample name	strength, σf (MPa)	elastic modulus, E (MPa)	fracture toughness, Gc (kJ m−2)
AM1 (PT, PF1)	1.45 ± 0.33	2.40 ± 0.46	1.30 ± 0.15
AM2 (PT, PF1)	1.63 ± 0.41	4.06 ± 0.99	1.83 ± 0.18
AM3 (PT, PF1)	1.34 ± 0.55	3.25 ± 1.70	0.96 ± 0.11
AM4 (PT, PF1)	2.00 ± 1.91	5.11 ± 1.55	1.69 ± 0.17
AM4 (TT, TF)	1.91 ± 0.66	7.37 ± 1.76	NA

Stress–strain curves in figure 2 also show the comparison of nonlinear responses between uniaxial and fracture tests. All membranes have similar stress–strain curve trends and fracture points for unnotched and pre-notched samples. Also, the stress–strain curves were similar regardless of different notch lengths. The stress–strain curves of tension samples (AM4 in figure 2f) exhibited slightly smaller stiffness than that of planar samples (AM4 in figure 2e), particularly at the beginning of the curves.

Samples with identical sizes were repeat tested for three pieces of amniotic membranes (AM1, AM2 and AM3). The failure parameters, including failure strain, failure stress and fracture toughness were consistent among PT and PF tests for each membrane (figure 3). The fracture toughness shown was determined from the measurement of both tensile and fracture tests following the method proposed by Rivlin & Thomas [21]. The standard variation is less than 15% showing the suitability of this method to quantify the toughness of AM.

Figure 3.

(a) Failure strain ɛ_f, (b) failure stress σ_f and (c) characteristic energy E_f between PT and PF tests as well as among four different pieces of amniotic membranes. The circles in (c) show the fracture toughness G_c of amniotic membranes. PF1 and PF2 show the fracture samples with different notch lengths. (Online version in colour.)

The characteristic energy E_f shown in figure 3 was determined from the measurement of a single type of test. The characteristic energy E_f for each test was similar to that of fracture toughness G_c but the standard variation is larger. All failure parameters of four amniotic membranes fall within the same range of deviations. It is generally accepted that the variation of measurement in mechanical properties of biological tissues tends to be larger than that in engineering materials. The variation resulted not only from the error introduced during the experiments but also due to the variation in the materials themselves, such as differences in membrane thickness and material microstructures.

3.2. Measurement consistency

The measurements were checked for the consistency of notch length, sample dimension and materials. For notch length study, the failure parameters, including the failure strain ɛ_f, failure stress σ_f, characteristic energy E_f and fracture toughness G_c, were plotted as a function of notch length (figure 4). All failure parameters also remained constant despite the different notch lengths introduced to the samples. Surprisingly the failure parameters of the unnotched samples of PT (a/w = 0) were similar to that of the pre-notched samples of PF tests (a/w > 0), indicating the extreme defect-tolerant characteristic of the tissue in the planar configuration. Similar failure stresses between the PT and the PF tests suggest that the material strength remains the same despite the presence of defects. The characteristic energy of the PT is similar to that of the PF tests and their value is within the range of variation of fracture toughness.

Figure 4.

(a) Failure strain ɛ_f, (b) failure stress σ_f, (c) characteristic energy E_f and (d) fracture toughness G_c of AM1 versus the length of notch. Insets indicate planar specimens (unnotched with a/w = 0) and fracture specimens (notched with a/w > 0).

For the sample geometry study, the failure parameters were plotted as a function of width-to-length ratio w/l (figure 5). A tension configuration is represented by (w/l < 1) while the planar configuration is represented by (w/l > 1). Two geometrical sizes of planar samples, Geo1 (w/l = 9) and Geo2 (w/l = 6), exhibited similar failure behaviour. The consistent measurement of fracture toughness G_c for these two dimensions validates that the failure criterion used in this study is independent of the sample dimension. The TF tests exhibited similar failure strain and stress to that of PF tests. However, it is unclear whether TF and PF tests have similar characteristic energy (figure 5); most measurements fall in the same range of deviation, but the average of characteristic energy of TF tests is larger than that of PF tests.

Figure 5.

(a) Failure strain ɛ_f, (b) failure stress σ_f, (c) characteristic energy E_f and (d) fracture toughness G_c versus geometrical sizes of specimens. The failure strain, strength and characteristic energy were taken from the measurement of unnotched specimens. Insets show the unnotched samples used in the measurement, consisting of TT (w/l < 1) and PT samples (w/l > 1).

4. Discussion

Nonlinear mechanical properties of human amniotic membranes have previously been characterized by uniaxial and biaxial tensile testing [4,10–12]. The stress–strain curves show a similar nonlinear J-shape, and the data are surprisingly insensitive to strain rate [17], and as such for testing the fracture behaviour here, the deformation rate was kept fixed. The tensile modulus of 7.37 ± 1.76 MPa is in the same range of the previous measurement of 6.80 ± 0.22 MPa [12]. The planar moduli obtained here have smaller values ranging from 2.40 ± 0.46 to 5.11 ± 1.55 MPa. The measured fracture toughness of four amniotic membranes from different mothers are 0.96 ± 0.11, 1.30 ± 0.15, 1.69 ± 0.17 and 1.83 ± 0.18 kJ m⁻². These values are comparable to the fracture toughness of other tissues obtained in mode III trouser tear test: 4.1 and 17 kJ m⁻² for adipose and dermal porcine tissues, respectively [22], 3.39 ± 0.57 to 5.40 ± 0.48 kJ m⁻² for cornea [23] and 1.21 ± 0.10 kJ m⁻² for skin [24].

The fracture toughness was determined based on the method purposed by Rivlin & Thomas [21], which has been tested in a rubber test piece. The unique feature different from the uniaxial tensile testing in previous studies is that the specimens were tested in pure shear condition. Each strip specimen was clamped along its long edges with a very small distance between the grips. Stretching of this specimen was approximate to pure shear deformation and is commonly used for the study of fracture mechanics of rubber [25]. The methods used to quantify the fracture toughness of soft tissues are relatively new and not well established. It is crucial for a mechanical property such as fracture toughness to remain a constant under different sample geometries and loading conditions. Many existing studies, however, overlook this important consideration [26]. In this paper, the measurement of fracture toughness has been checked to be consistent in specimens with different geometries and notch lengths. It was found that the measurement of failure strain and fracture energy was also independent of crack length in planar specimens. Rivlin & Thomas recommended that the width of the pure shear test piece must be at least four times the initial length. The tensile samples tested in this study, which had initial lengths longer than their width, showed the inconsistent characteristic energy (figure 5c). By contrast, the planar samples having width more than four times the sample length provide consistent characteristic energy and fracture parameters. The measurement consistency supports the suitability of Rivlin & Thomas method to quantify the fracture toughness of soft tissues such as amniotic membranes.

Despite the intention to study how the defects in membranes induce rupture, we found that amnion is very tolerant and insensitive to the existence of defects. Failure parameters including material strength and failure strain remained approximately the same despite the presence of a notch. Surprisingly the stress–strain curves of the pre-notched and the unnotched specimen were also found similar. By having similar stress–strain curves, the area beneath the stress–strain curve integrated up to the fracture point w_f for an individual sample is similar to the fracture energy density w_o determined according to Rivlin–Thomas method. A characteristic energy E_f presented in this paper can be used as a fracture criterion for the toughness of extreme defect-tolerant amniotic membranes.

The flaw tolerance can be explained by the deformation and toughening mechanism inherent in the hierarchical microstructure of amniotic membranes at the blunting crack tip. Extreme fracture resistance characteristic has also been observed in other soft tissues like skin [27]. A notch in the skin was found not propagating but only opening and blunting. The examination on the microstructures showed mechanisms of fibril straightening, fibril reorientation, elastic stretching and interfibrillar sliding release energy and redistribution of the stresses at the notch tip in the skin [27]. Mechanisms of rearrangement of fibrous networks to form bundles [10,11] and uncrimping of collagen fibres [28] were observed in the deformation of amniotic membranes. The toughening mechanisms have been also explained by the formation of fibre bundles in engineering fibrous materials in both experiments [29] and simulation [30]. This study does not consider some features that potentially affect membrane rupture during PROM or PPROM. For instance, the time-dependent behaviours that affect membrane behaviour, membrane thickness and the integrity of amnion–chorion interface due to water outflow [4,10,11,16,31].

The inherent toughness and extreme defect-tolerant characteristics of amniotic membranes observed in this study can possibly explain why some preterm labours with pathologic complications experience no rupture of membranes [32]. The inflammation and infection of the membranes can cause the formation of a flaw, which has altered morphology and poorer mechanical properties. Such a flaw is comparable to the notch introduced in specimens in fracture testing here. The fracturing of notched specimens in this work showed extreme tolerance to the existence of flaws in the membrane, and the large fracture toughness values observed further demonstrate the ability of the membrane to resist rupture from the flawed region. Questions have been raised as to whether membrane rupture is dominantly caused by and initiated from the weak zone in the amnion. While the excellent toughness properties of membranes can decelerate the rupturing of membranes from mechanical forces, the progressive involvement of biochemical processes possibly plays an important role in triggering the expansion of the area of the weak region that eventually causes PROM and PPROM [7].

5. Conclusion

This work is the first study on the measurement of fracture toughness of amniotic membranes. Uniaxial tensile tests and fracture tests were performed on human amniotic membranes in order to quantify the mechanical condition that will fracture the membrane with defects. The fracture toughness of amniotic membranes has been measured and the measurement is consistent across sample geometry. The results show that amniotic membranes are tough and tolerant to defects. The results in this study can provide an understanding of membrane rupture occurring before deliveries both at term and preterm.

Ethics

Ethical approval for the study was given by the Cambridgeshire 2 Research Ethics Committee (reference number 07/H0308/163). All participants gave informed written consent.

Data accessibility

All raw and analysed experimental data are shown herein.

Authors' contributions

The study was conceived of jointly by C.T.K. and M.L.O.; the experimental and analysis work was performed by C.T.K. as part of her PhD thesis supervised by and in discussions with M.L.O. The experiments were conducted with contributing collaboration from K.T. The manuscript was a joint effort between C.T.K. and M.L.O.

Competing interests

We declare we have no competing interests.

Funding

Funding for this study was provided by the Ministry of Higher Education Malaysia via a PhD studentship to C.T.K.