Standardized Bending and Breaking Test for the Normal and Osteoporotic Metaphyseal Tibias of the Rat: Effect of Estradiol, Testosterone, and Raloxifene


  • The authors have no conflict of interest


The fracture of bone plays a key role in osteoporosis. BMD measurement, however, is only an indirect parameter of this phenomenon. We therefore developed a highly sensitive three-point bending test for the metaphyseal tibias in rats to evaluate stiffness and strength. This was validated in a right-left comparison and a bioassay with soy-free food, estradiol, raloxifene, and testosterone in orchidectomized rats.

Introduction: Osteoporosis becomes manifest predominantly in the metaphyseal rat tibia. The anti-osteoporotic character of substances should, therefore, be tested (mechanically) in this bone area.

Materials and Methods: We evaluated a new three-point bending test for the metaphyseal tibia in rats in a right-left trial. In an animal experiment, we studied the change of bone quality under estradiol (E)-, raloxifene (R)-, and testosterone (T)-supplemented food and compared it with trabecular BMD (qCT).

Results: In the right-left comparison, the mean difference between the metaphyseal loads of both tibias in 37 rats was 8.43% for the maximum load (Fmax) and 6.46% for the failure load (fL). These results show the high reproducibility of the test, because they are close to the usual intraindividual difference of the two extremities. In a second experiment, four groups of 11 3-month-old male orchidectomized rats were fed with soy-free food only (C) or with the additives E, T, or R for 12 weeks. E and R were similar for Fmax and fL. There were significant differences in the stiffness (E = 406.92 N/mm versus R = 332.08 N/mm), the yield load (yL; E = 99.17 N versus R = 83.33 N), and the ratio between yL and Fmax (E = 86.33% versus R = 76.37%). T was similar to the controls concerning Fmax, fL, and stiffness. There were significant differences in yL (T = 49.00N versus C = 39.5N) and the ratio between yL and Fmax (T = 64.28% versus C = 51.28%).

Conclusions: Estradiol is superior to raloxifene concerning stiffness and yield load, and both are superior to testosterone. We conclude that the described three-point bending test for the metaphyseal tibia is a highly sensitive method to study hormones and substances with regard to their osteoprotective character. The precision and the low SD of the presented results are superior to the data from qCT and the calculated index of stiffness (SSI).


THERE ARE NUMEROUS causes of osteoporosis, the most frequent being lack of estrogen. About 30% of all postmenopausal women develop osteoporosis within 10–15 years of menopause. (1, 2) The annual costs caused by osteoporotic fractures in 2003 in the United States were estimated at up to $16.7 billion. (3) In Germany, the entire costs for osteoporotic therapy amount to 10 billion Euro a year. (4) Furthermore, osteoporotic bone heals ∼30% more slowly(5, 6) than normal bone.

Hormone deficiency leads to a reduction of bone structure, and even minor trauma can cause fractures. These fractures are predominantly located in the metaphysis of long bones (i.e., the distal radius, the proximal femur, and the lumbar spine). BMD measurements to evaluate the degree of osteoporosis are therefore carried out at the femoral neck and the lumbar spine.

The ovariectomized rat is an established model of osteopenia and osteoporosis. (7–9) Animals develop substantial osteoporosis within a few months of ovariectomy such that, after 6 months, ∼50% of bone mass has been lost. (10, 11) From previous studies it is well known that the degree of osteoporosis is best determined at the metaphyseal femur or tibia of the rat, (11, 12) trabecular density being the key parameter. The diaphysis of long bones consists of cortex and bone marrow. The growth plate does not close during the lifetime of the rat. Osteoporotic changes do not manifest in the diaphysis because there is no trabecular bone. The diameter of the diaphysis can increase only slightly. Nevertheless, most of the experimental mechanical tests to evaluate osteoporosis or osteoprotective substance have been carried out in the diaphysis of the femur or tibia. (13, 14) In this area of the long bones, both a three-point bending test and even a four-point bending test are possible. These are easier to manage than a standardized metaphyseal fracture. In contrast to metaphyseal BMD or histomorphometrical measurements determined in the same animal and time, (15, 16) the results of the diaphyseal bending or break tests are not able to reveal significant differences in osteoporosis.

We postulate that mechanical testing of the metaphysis of long bones is necessary to evaluate the special properties of osteoporotic bone. BMD measurement is an indirect parameter only, which provides poor evidence for the real stiffness and actual resistance to applied force. Low BMD without a fracture would not be a problem. The manifest fracture is the crucial point in osteoporosis. In the osteopenia model of the rat, the metaphyseal tibia is the most reproducible fracture area.


Animals and substances

The primary evaluation of the newly developed three-point break test was carried out with rats that were at least 3 months old.

Right and left tibias of 37 rats differing in age, strain (Sprague-Dawley rats, Wistar rats, etc.), and sex were studied to show that the three-point break test is universally applicable.

In the second part of this study, 44 3-month-old male Sprague-Dawley rats (220-260 g) were orchidectomized and divided into four groups. The first group received phytoestrogen-free pelleted food (protein supplementation was secured by added potato proteins). The second group received phytoestrogen-free food supplemented with estradiol benzoate (E). Average food intake per animal per day was 20 g, so that the average estrogen intake was 0.5 mg E/day, corresponding to 0.325 mg free estradiol-17β. The third group received phytoestrogen-free food supplemented with testosterone propionate (T), so that the average testosterone intake was 50 mg/day, corresponding to 35 mg free T; the fourth group food was supplemented with raloxifene (R) at an average intake of 3.35 mg R/day. Animals were kept on this food for 12 weeks. The study protocol was approved by the District Government and conformed to German animal protection laws (permission from 11.03.1998, Az: 509.42502/01-02.98 Bezirksregierung Braunschweig). In one tibia, we tested the breaking strength, whereas the other tibia was used for immunohistochemical studies published elsewhere. (17, 18)

Animals were killed under anesthesia. Both tibias were prepared as follows: skin, muscles, and tendons were removed, and the fibula was separated at the synostosis. We also removed the proximal tibia epiphysis because we found a luxation of the proximal growth plate during the breaking test in preliminary experiments. The growth plate of the tibia does not close in the rat. Tibias were stored at −20°C until use.

Trabecular density

Animals were anesthetized (inhalation anesthesia with isoflurane) 12 weeks after onset of E-, T-, or R-supplemented food, and the BMD of the metaphysis of the tibia was recorded by qCT (XCT 2000 Research; Stratec Medizintechnik, Pforzheim, Germany). Using the software, an index of stiffness (SSI), representing the geometrical properties of the bone, was calculated. Properties such as failure load or elasticity cannot be calculated from qCT data.

Serum analyses

Blood samples collected from animals were allowed to clot and were then centrifuged (3000g, 10 minutes). Serum concentrations of estradiol-17β and testosterone were quantified with commercial immunoassays (E: Elecsys; Roche, Mannheim, Germany; T: DSL; Webster, TX, USA).

Mechanical testing set

We used the ZWICK-testing machine type 145660 Z020/TND (Zwick/Roell, Ulm, Germany). The measuring range was from 2N to 200N at a relative accuracy of 0.2% at 0.4% nominal force (FN). We chose a primary force of 1N to fix the tibia on the device. Strength admission was recorded using “testXpert” software. The speed of the feed motion was 50 mm/s, and the automatic switch off-pressure was set at 300N. The trial was automatically ended by a drop in strength of >20N or a linear displacement of >2 mm—this corresponds with the fracture of the complete cortical bone—to avoid bursting of the tibia.

Development of the three-point break test

A four-point bending or breaking test is standard in large animals such as dogs, pigs, or sheep. (19–21) However, the metaphyseal length of the rat tibia is too small for this test; the entire tibia has a length of 20–35 mm, with a diameter of 4–6 mm at the condyles. The aim of our project was to conceive a standardized metaphyseal fracture of the tibia in rats to evaluate the degree of osteoporosis. We therefore developed a three-point breaking test, which is mechanically similar to the four-point bending tests in larger animals.

The stamp and the base, where the tibia was positioned, had to be optimized in several experiments (material, size, shape) to minimize friction between the bone and the base. Finally, a base consisting of an aluminium block (20 mm wide, 12 mm high, 80 mm long) with three rounded edge-free notches (1 mm deep; 2, 3, or 4 mm in diameter) on top was found to be optimal. Depending on the size of the tibia, the end of the dorsal proximal tibia (the two condyles) was placed in one of the notches. Because of its S shape, the tibia rested on the base with the two condyles in the notch and with the distal diaphysis at the (former) synostosis with the fibula on the other side. This position of the tibia resulted in stable three-point contact with the base (Fig. 1A). During the breaking test, the tibia could not slip, but it was able to lengthen along the diaphyseal axis. The tip of the stamp consisted of an axle-led aluminium roller (8 mm high, 8 mm wide, 6 mm in diameter). A 2-mm-wide and 1-mm-deep circular notch with rounded edges was located in the center of the roller. Because of the mobility of the stamp strength loss, lateral slipping and a point crush into the bone could be prevented. The roller axle was fixed in a U-shaped support, which was connected to the ZWICK-test machine by an aluminium stem.

Figure FIG. 1..

(A) Details of the three-point bending test consisting of the ZWICK-testing machine (measuring range from 2N to 200N), the aluminium base, and the roller stamp (drawing available on request). Tibia in position before starting the test. (B) Radiographs showing the typical metaphyseal tibia fracture.20

Mechanical testing procedure

The tibias were thawed and continuously moistened with isotonic saline solution. Each tibia was placed with the three-point contact on the aluminium base. The base was fixed in the ZWICK-testing set with a distance of exactly 3 mm between the end of the proximal tibia (without the epiphysis) and the center of the roller stamp. The stamp was driven down to the ventral metaphysis of the tibia till the primary strength of 1N was reached. After a final visual check of the correct tibia position, the breaking test was started. The study ended automatically by a loss of strength of >20N or a linear change of >2 mm. The procedure was done blinded with regard to the test groups.

Radiographic evaluation

Radiographs in the anterior-posterior and the lateral view were taken of all tibias from the comparative bioassay. A special film (Fuji HR-E 30 Medical X-ray) and a Faxitron Cabinet X-ray system (Model 43855A; Faxitron X-ray System) with 40 KV were used. The exposure time was 6 minutes. The individual responsible for the radiographic evaluation and for describing the fracture type was also blinded to the test groups.

Evaluation and statistics

During the breaking test, the actual strength was recorded every 0.1 mm during the lowering of the stamp. The testXpert software continuously recorded the force applied until the total failure of the bone occurred. After the failure, the software program indicated the maximum load (Fmax), the area below the graph, and the breaking strength. Because of fluctuations of the graphs during the test—probably because of consolidation—the energy to failure was sometimes visible inaccurately, because it corresponds to the area below the graph. This is a phenomenon of the bone and not of the developed breaking test. We did not evaluate it in the animal experiments. The breaking strength is the last measured point of the running graph; it has no explanatory power. In the right-left comparison and the comparative bioassay, we took the penultimate strength value to bone failure instead and called it failure load (fL). Fmax is the highest strength that the tibia can withstand.

From the digital data, the rise of the elastic deformation (stiffness) was calculated, and the transition point of elastic to plastic deformation was determined. This point represents the yield load (yL) of the bone. To determine this point, we calculated a regression line and the SD with the individual data of the linear part of the graph. We defined the transition point of elastic to plastic deformation as a decrease of stiffness of more than twice the SD. As an indicator of bone quality, we calculated the ratio of the force at the yL and Fmax.

The results of right-left comparison in 37 rats were statistically analyzed in an unconnected pair test. The proportional difference between the strengths of the right and the left tibia was determined in each rat, and the average value was calculated. The average value of the proportional differences of the maximum load and the failure load is a sign of the reproducibility and the quality of our new three-point breaking test. In the comparative bioassay, 11 rats per group were evaluated and compared. Differences between the groups were assessed using a paired t-test.


Graph of the application of strength

A typical graph of the application of strength to the tibia can be divided into three parts (Fig. 2). In the first part, it rises linearly. This describes the elastic deformation of the bone. In the second part, the slope declines until the maximum load is reached. This is the first plastic deformation with microfracturing of trabecular bone. It starts with the (calculated) yL and ends with Fmax. In the last part, the graph declines because of multiple fracturing until complete failure of the bone occurs, the metaphyseal fL. The machine automatically stops the breaking test.

Figure FIG. 2..

Typical graph of the application of strength (raloxifene group): In the first part, there is a linear increase until the first trabecular fractures occur (yield load). In the second part, plastic deformation occurs as a result of microfracturing. In the third part, the maximum load (Fmax) is reached, and the graph drops slowly until the cortical bone fails (failure load).20

Right-left comparison

Both tibias from 37 rats of different strains were studied with our breaking test. The maximum load (Fmax) showed an average difference of 6.46%, whereas the fL showed a difference of 8.43%, and the energy to failure showed a difference of 16.3% between the right and left tibias. The weight of the animals ranged from 178 to 544 g. Accordingly, the absolute values of yL, stiffness, Fmax, fL, and energy to failure are not comparable: they differ between 16 ± 1N and 148 ± 13N depending on the size, age, and sex of the rats (Fig. 3A).

Figure FIG. 3..

Scatter plots of the results of the right-left comparison. (A) Stiffness (S). (B) Failure load (fL). (C) Yield load (yL). (D) Maximum load (Fmax). (E) Energy to failure (Ef). (F) Two graphs of the right-left comparison in one rat showing the parallelism of the graphs. The differences between the Fmax and failure load of the two curves were 7.9% and 9.5%, respectively.20

The comparison between the two graphs of strength of both tibias in one rat showed great similarity (Fig. 3F). Fluctuations of the graphs sometimes occurred during the test, probably because of consolidation; however, they did not influence the yL, stiffness, Fmax, or fL, but they did influence the energy to failure.

Effects of estradiol, testosterone, and raloxifene

After orchidectomy, 11 rats per group were examined. The right tibia was used in the breaking test, and the left was examined in immunohistochemical and molecular biological tests. (17, 18) Table 1 shows the results; Table 2 shows the significances.

Table Table 1.. Results of the Comparative Bioassay of the Four Substances
original image
Table Table 2.. Significance Between the Results of Treatment in the Comparative Bioassay
original image

Classification of fracture in the comparative bioassay

Three types of fracture could be differentiated: the metaphyseal oblique fracture, the Y-fracture, and the unilateral short oblique or condyle fracture. In the four groups of the comparative bioassay, the last type was dominant (six each in the C and E groups, seven in the T group, and five in the R group). The Y-fracture occurred in only 2 of 11 rats in the control, 4 in the E, 1 in the T, and 3 in the R group. In the remaining animals, we found a metaphyseal oblique fracture. We did not see similarities in one group, nor did we find feed-specific differences between the four groups.

Serum analyses

Serum concentrations of estradiol-17β in the E-treated animals were 51.6 ± 4.6 pg/ml, and in the androgen-fed animals, T levels were 4.2 ± 0.8 ng/ml.

Trabecular density and SSI

Three months after castration, the rats showed >30% loss of BMD (72.02 ± 19.42%) compared with the values before castration (=100%). R (96.35 ± 24.18%) nearly compensated the effect of castration on bone. E (86.78 ± 11.46%) and T (80.95 ± 16.7%) only partially prevented this loss of BMD of the metaphysis of the tibia. The values from the treated animals showed significant differences to the control group but not between the three tested substances themselves. The calculated index of stiffness (SSI polar) was 15.19 ± 4.51 mm3 for the control group, 16.8 ± 5.58 mm3 for the T group, 15.93 ± 6.22 mm3 for the R group, and 13.23 ± 5.41 mm3 for the E group. There were no significant differences.


Serum analyses

In this study, we showed that substitution of testosterone in orchidectomized rats or treatment with estradiol had profound effects on the bone. Serum E concentrations were in the range considered physiologic for females. The serum T concentrations achieved in the animals receiving the T-containing food were in the physiologic range. Hence, T substitution after orchidectomy was adequate.

Standardized metaphyseal tibia fracture

The success of osteoprotective therapy is often examined by immunohistochemical studies of osteoblastic growth factors and gene expression and measurement of BMD(17, 18) by histomorphometric studies or mechanical testing. (7, 13, 14, 16, 22) Usually the mechanical testing is done at the diaphyseal femur or tibia. However, osteoporosis primarily affects the trabecular bone in the metaphysis of long bones in rats and humans. In the rat, the metaphysis of the tibia is particularly predisposed to develop osteoporosis. Therefore, we chose this region for mechanical testing and were able to show changes in stiffness, plastic deformation, and breaking strength. Some authors have studied the mechanical properties of bone in osteoporotic isolated vertebral bodies (removed endplates and/or spinal processes) or body cylinders, (13, 23–25) which yielded no information about the intact vertebral body. Our experimental breaking device allowed evaluation of mechanical properties of intact bones.

The size of the rat tibias we studied in the right-left comparison varied between 20.3 and 32.2 mm. We therefore consider a three-point breaking test as convenient. To prevent slipping or tipping of the tibia under increasing punctual strength, we developed the stamp for the testing machine as a roller with a central notch (Fig. 1B). The results of the right-left comparison underline the reliability of our three-point bending and breaking test, because of the minimal intraindividual deviation.

Right-left comparison

The intraindividual deviation of fL and Fmax (∼6% and 8%) corresponds with the dimensions of the physiological difference between the right and left extremity. Side comparative studies in human and animals show intraindividual variations of the same dimension that we found in our right-left comparison. (26–29) In humans, we know there are differences in size and strength of the right and left lower extremity (anklebone) or upper extremity (right-handed or left-handed person).

Breaking strength under estradiol, testosterone, or raloxifene

The analysis of the graph of the breaking test allows conclusions regarding the mechanical properties of normal and osteoporotic bone. The linear part of the graph represents the stiffness of bone. No microfracturing or trabecular fractures occur during this elastic elongation. This is of special interest if nondestructive mechanical testing is necessary (e.g., microscopic studies of bone structure may be disturbed by fractures). With increasing force, the first microfractures occur. This is the point of change from elastic to plastic deformation: the yL of the bone. During microfracturing, the bone is able to bear more force up to the maximum (Fmax) when the sum of microfractures decimate it too much. Finally, the bone breaks completely (fL).

There was a correlation between the yL and the Fmax of a bone. This ratio should be high in normal bone and low in osteoporotic bone, because normal bone is highly elastic, whereas osteoporotic bones have trabecular microfracturing at a very early stage.

In the animal experiment of our breaking test of the tibia metaphysis, there was no (significant) difference between the E group and the R group concerning maximum load and failure load (Table 1; Fig. 4). This correlates with the known effects of both substances. (30, 31) There was a significant difference in favor of E compared with R in the stiffness (rise of the elastic deformation; 406.92 versus 332.08 N/mm) and in the yield load (99.17N versus 83.33N) in the presented data. These differences were confirmed by the ratio of yL and Fmax: E (86.33%) differed significantly from R (76.37%). These data suggest that the yL of bone is significantly higher in animals treated with E compared with R. This means that there will be the same fracture protection with E and R at a unique higher force (e.g., a fall), whereas lower forces will lead to earlier microfracturing with R than with E. Multiple minor traumas may lead to creep fractures. This creep process is accompanied by the progressive accumulation of microfissures within the bone, especially trabecular bone, (32) which could cause earlier fractures with R.

Figure FIG. 4..

Results of the comparative bioassay. (A) Change of the bone stiffness under testosterone, estradiol, raloxifene, or soyfree food. (B) Effect of the different treatment on bone deformation (yield load), maximum load, and the failure load. (average values and SD, significances in Table 2).20

The results after application of testosterone differ from the other data. (33–35) We found no difference in the maximum load and the failure load in the T group compared with the orchidectomized controls. The linear rise of the graph, representing the stiffness of the bone, also showed little difference. However, the mean yL of bone treated with T was significantly higher than that of the controls (49.0N versus 39.5N). The ratio of yL to the maximum load was also significantly higher in the T group compared with controls (64.28% versus 51.28%). We presume that the stiffness of bone is similar after testosterone therapy compared with the controls, but plastic deformation occurs significantly earlier in the control group. Thus, there will be higher stress protection similar to that observed with raloxifene.

The maximum bone strengths in the E and R groups were significantly higher than in the T group and the orchidectomized controls. However, there was a significant gradient in the stiffness of bone and yL: E - R - T - C (Table 1). We consider that the first plastic deformation is the start of the trabecular microfracturing of cancellous bone. The maximum load is the last point at which the bone can withstand the increasing force despite microfracturing. The moment of complete failure (fL) represents the cortical fracture.

Trabecular density, SSI, and breaking strength

The trabecular BMDs of animals from the E, R, and T groups measured with qCT were significantly higher than in the soy-free controls. Because of the large SD, there were no significant differences between the three tested substances. In contrast, the SD of the new bending test was notably lower. In the bending test, we observed significant differences between T and E or R for Fmax, fL, and stiffness. At yL, the point of elastic to plastic deformation, there were even significant differences between all tested substances.

In terms of their dimension, independent of the significance, the results of the mechanical testing and the values of the trabecular density were related, but the calculated index of stiffness (SSI) showed neither a relationship with the trabecular density nor with the mechanical testing results. We propose that it is better to directly study the biomechanical properties of bone rather than to calculate or to deduce them solely from qCT data.

The biomechanical properties of bone are primarily determined by the elastic properties of the extracellular matrix. Both qCT and microradiography indirectly measure these properties through the absorption of X-rays. The geometrical structure of bone in the 3D space cannot be differentiated with qCT. For example, the stability of a house does not only correlate with the thickness of its walls, but it also depends on their quality (i.e., walls of reinforced concrete or bricks).

In clinical practice, noninvasive biomechanical measurement of bone quality for diagnostics or therapy control is not possible, and it is well known that the real individual fracture risk cannot be measured by BMD. In the future, it will be necessary to study the extracellular bone matrix with regard to the mechanical properties of bone and the changes of bone during developing osteoporosis.

Type of tibia fracture

The analysis of the different types of fractures in the animal experiments showed neither similarity in one group nor clear differences between the four groups. In all groups, the unilateral short oblique or condyle fracture was the most frequent type. The radiographs allow no conclusion of changes of bone quality corresponding to the treatment.

In summary, the new metaphyseal three-point breaking test could be standardized in the right-left comparison of the tibia in rats. Significant differences could be revealed in an animal model between the orchidectomized control, E, R, and T groups. The SD of the results was low. In summary, we postulate that our three-point breaking test is highly sensitive to evaluate hormones, substances, and even medications with regard to their osteoprotective efficiency, because the osteoporotic fracture plays the key role in osteoporosis.

In the comparison of three groups with osteoprotective therapy, the point of change from the elastic to the plastic bone deformation, yL, gives us valuable new information about bone failure and fracture. The yL reveals significant differences between the four groups, which could not be detected by measuring the maximum load (Fmax) of bone only. We conclude that E and R are able to protect cortical bone from osteoporotic fractures. Concerning the stiffness and the strength of the cancellous bone, E seems to be superior to R, and both of them are superior to T, but T is still superior to the controls.


The authors thank Dr VW Armstrong (Department of Clinical Chemistry) for lingual help.