Femur sections

The right femurs were thawed, and their weight and length determined after any residual adherent soft tissue had been removed. The measured length of each femur was then divided into five segments marked at 25%, 45%, 65% and 85% of the total length. Each segment was then sectioned along the transverse lines with a diamond wheel saw. These segments spanned 60% of the overall length, and extended from the upper limit of the intercondylar ridge (segment C) at the distal end of the femoral shaft, to the upper limit of the third trochanter (segment A) at the proximal end of the femoral shaft. Each section was marked at 50% of its length and stored in PBS at 4° C. All segments were checked to ensure that they had not been damaged during cutting. Any cracked

segments were discarded.

6.2.3. Izod test

The Izod test is an impact test traditionally used in engineering as a comparative tool to measure the resistance to failure of a material to a suddenly applied force (13). This is expressed as impact energy, or the amount of energy needed to fracture the test sample. The pendulum has two types of energy - potential and kinetic. The maximum potential energy is stored at the start of the swing and converts to maximum kinetic energy at the lowest point of the swing (Fig. 6-1). The striker on the pendulum impacts the bone sample at the moment of maximum kinetic energy, at this point the energy is absorbed by the bone until fractures occurs. The pendulum then continues on into an upward swing. The impact energy (Joules) is calculated by the loss of energy in the pendulum swing after it strikes the bone sample (7, 13). The difference represents how much

6-5 energy was absorbed by the bone sample before it fractured. A tougher bone sample will absorb more energy before fracturing compared to a brittle bone sample, which will fracture with less energy. The specific energy (Joules/mm2) of a material is the amount of energy per unit of area required to fracture that material.

The above scenario assumes that the pendulum pivot is frictionless and there is no air resistance on the striker as it swings. Therefore, to reduce any possible error that these assumptions make, the pendulum was let freefall 10 times before testing began, and after every ten bone segments tested. The height of the pendulum was recorded after every freefall swing and the average energy lost by the pendulum was subtracted off the results of each segment.

Figure 6-1. Diagram of Izod test.

The biomechanical testing was done using a “Zwich” impact testing machine

(Zwick/Roell, Germany). Each bone sample was gripped firmly at the distal end of the segment, leaving the mid and proximal portions of the segment exposed (Fig. 6-2a). The bone samples were gripped at a height to ensure that the striker impacted the bone segment consistently at 50% the length of the sample. All segments were held under the same orientation to reduce error, and kept wet throughout the test.

6-6 The pendulum had 0.5 Joules of potential energy at the start of each swing. It was held at a 160° angle and let freefall until it hit the sample. Upon impact the bone segment was broken completely into several pieces. The impact energy (J) lost from the pendulum to the bone segment across the bone segments cross section (mm2) was recorded and expressed as energy absorbed per unit of bone area (J/mm2) (8). This is the material’s specific energy. The bone area measurement was taken at the point of impact across the transverse cross section of the bone (Fig. 6-2b). The bone area of the samples could not be taken directly from the bone segments broken in the impact test, as they were too damaged to accurately determine the area. Therefore the measurements were taken from the left femur bone area measurements, under the assumption that differences in bone size and bone area between left and right femurs are not significant (14).

6-7

b) Diagram of bone sample showing the point of impact between the pendulum and the bone segment, with the impact occurring half way up the length of the segment. The cross section

shows the area of the bone used to calculate the bone area (mm2_{) measurement.}

Figure 6-2. Diagrams showing orientation of bone samples in Izod tester. 6.2.4. Statistical analysis

It was decided to run a separate analysis of the rats fed the non-milk diet from the rats fed the milk diets. A first attempt to run the data together failed as the data from the non-milk groups prevented the data from normalizing. Therefore the analysis of the milk diets, GOAT OVX, GOAT OVX ALD, COW OVX and COW OVX ALD were run as a separate series of analyses to the rats on the non-milk diet (SHAM, OVX and OVX ALD).

Data were analysed in the statistical package “SYSTAT” (version 11) (Systat, Chicago, USA). Statistical significance was set at p<0.05. Biomechanical measurements were found for the most part to be normally distributed for the segments. Where non-normal data were found a log transformation was required to obtain near normal distribution on graphic analysis. Impact energy (J) for segments A and B were transformed using a Johnson transformation (15), as was bone area (mm2) for segments A and B with the statistical package “MINITAB” (Minitab Inc, Pennsylvania, USA). Outliers were removed from some groups to allow for data to be transformed to a normal distribution. The number of outliers removed never exceeded 10% of the group number. The data for segments A and B were then transformed using

Y = 1.198 + 0.950 * [ln(x-0.23) / (0.214-x)] Similarly bone areas from segments A and B were transformed using

6-8 Y = 0.325 + 0.958 * [ln(x-5.681) / (10.60-x)]

Specific energy (J/mm2) for segment C data, and Impact energy (J), could not be transformed to represent a normal distribution and all data for this segment were therefore excluded from further analysis. This was likely a result of extensive in- homogeneity of transverse slices in the segment.

Groups fed the Goat or Cow’s milk diets

The variation of impact energy, with treatments was assessed by 2-way ANOVA. The relationships between the impact energy (J) used to break the bone samples and the bone area (mm2), and total cross sectional area (mm2) of the proximal and mid-segments were assessed by ANCOVA’s and linear regressions. Student’s T tests were used to compare the slope of the linear regression and were calculated by,

T = (a1-a2)/SEa1-a2

where, a1 = mean slope of the ovariectomized + Alendronate treatment group and a2= mean slope of the ovariectomized group. SE = standard error of the comparison, and was calculated using

SEa1-a2 = √ SEa12– SEa22 (16).

Statistical significance of T was intercalated from the students T table.

The mean overall bone area was calculated from the combined areas of the four slices of the left femur for each of the segments A and B. This was repeated for overall cross sectional area. The variation of bone area and overall cross sectional area, with treatments was assessed by 2-way ANOVA.

Groups fed the Non-milk diet

The impact energy of the three groups of rats fed the non-milk diet was examined by one-way ANOVA. The relationships between the impact energy (J) used to break the bone samples and the bone area (mm2), and total cross sectional area (mm2) of the proximal and mid-segments were assessed by ANCOVA’s and linear regressions. Student’s T tests were used to compare the slope of the linear regression (see above for equations).

6-9 The mean overall bone area was calculated from the combined areas of the four slices of the left femur for each of the segments A and B. The variation of bone area and overall cross sectional area, with treatments was assessed by one-way ANOVA and significant differences identified where examined using Tukey post-hoc testing.

6.3.

Results

6.3.1. Milk diets – Segment A