All experimental rabbits were fed in a single cage and received a standard laboratory diet. Experimental protocols and animal care methods were approved by the Animal Experimentation Ethics Committee of the Stem Cell Institute, University of Science, Vietnam National University, Ho Chi Minh City, Vietnam.

Male New Zealand white rabbits (28 weeks of age), weighing 2.5-3.5 kg, were separated into two groups. The ONFH group (n=10) was injected in the gluteus medius site with a dose of 0.5 mg/kg CFA (Santa Cruz, USA) at day 1 and 40 mg/ kg MPS (Pfizer, New York city, NY) at day 7. The normal group (n=6) received 0.9% saline at the same location and volume as those in the ONFH group. Rabbits were assessed at week 7 after the last injection. All rabbits were followed to record any body weight changes.

Seven weeks later, rabbits were sacrificed. Femoral head and thighbone were collected. The morphology was observed and recorded using a camera.

In the next step, both femoral head and thighbone were fixed in 10% paraformaldehyde for 3 days. After that they were soaked in 10% disodium ethylene diamine triacetic acid (Na2-EDTA) to decalcify. Na2-EDTA solution was refreshed every 2 days until sufficient demineralization was achieved. Demineralization was detected by stabbing a needle to the bone and observing the surface color change of samples. The samples were then dehydrated in 30% sucrose solution. After 2 days, the samples were cut into 7-10 µm-thick sections and stained with hematoxylin-eosin and safranin O. The changes in the histological structure of the periosteum, trabeculae and cartilage were observed using a microscope. Ten random fields were chosen to count the number of empty osteocyte lacunae. Bone damage level was detected by the rate of empty osteocyte lacunae.

Ink artery infusion angiography was conducted according to the method of Wen Qian et al. Wen et al., 2008b. Briefly, the rabbit was anesthetized and injected with heparin (5,000 IU/mL) into an ear vein. The abdomen was operated to expose the abdominal aorta and the abdominal vein was cut. A 12-gauge syringe was inserted in the opposite direction of the heart. About 2,000 mL of physiological saline solution was pumped to wash away the blood from the body. Then, about 100 mL of gelatin and ink mixture (at a rate of 7:3 in phosphate-buffered saline) were injected continuously until the skin and nails became uniformly black. Viscera from the rabbit was removed and kept at 4°C overnight. Over the next day, the femur was collected and fixed in 10% paraformaldehyde for 3 days. Subsequently, samples were decalcified as described above and cut into 30 µm-thick sections and blood vessel density was observed using a microscope.

To assess the effect of CFA and MPS on the proliferation and apoptosis of nuclear cells in bone marrow, rabbits received epidural anesthesia using 3% Novocain (Hanvet, Ho Chi Minh city, Vietnam). The bone marrow was harvested through a hole in the top of the ilium with a 18-gauge biopsy needle. Bone marrow-derived nuclear cells were enriched by incubation in hemolysis solution. Nuclear cells were fixed in 70% ethanol for 24 hours and stained with propidium iodide (Sigma-Aldrich, Louis St, MO) for 10 minutes. Cell cycle was analyzed by flow cytometry.

All data were analyzed using Graphpad Prism 6.0 software. Images were analyzed using ImageJ software.

After 7 weeks of drug administration, bone morphology had significant changes in normal rabbits compared to treated rabbits ( Figure 1 ). The femoral head was shiny, smooth and opalescent in the normal group. In contrast, in the treatment group, the cartilage surface of the femoral head was slightly rough, uneven and pink.

In group B, almost all osteocyte lacunae were filled with cells that was identified by the presence of oval-shaped nuclei inside ( Figure 2B ). In contrast, the number of empty lacunae was displayed at many bone sites in the ONFH group. The percentage of empty osteocyte lacunae was significantly higher in ONFH group than normal group (p <0.05). The percentage of empty bone lacunae defects in the normal group was 0.06 ± 0.01%, while for the ONFH group it was 0.17 ± 0.02% ( Figure 2 ).

The concentration of proteoglycans that displayed by orange color was significant higher in the normal group ( Figure 2A,B ), compared to the ONFH group ( Figure 3A,D ) at 7 weeks. Statistical data showed that the proteoglycan level of the control group was 4.14 ± 0.29 μm while for the ONFH group the values were 2.29 ± 0.17 μm (p <0.05).

At the zone of calcified and hypertrophic cartilage, the fibrocartilage (characterized by the presentation of collagen via green color) was strongly expressed in the ONFH group after staining with safranin O and fast green (narrow; Figure 3D ), as opposed to the control group (narrow, Figure 3B ).

The capillaries network in the medullary cavity was high at femoral heads of the normal group, as displayed by the presentation of black color after ink injection into the abdominal aorta ( Figure 4A ). In contrast, in the ONFH model group, the density of black medullary cavity was less, indicating that the capillary network was sparse and that there might have been some injuries within the femoral head ( Figure 4B ).

Compared to the control group, the proportion of bone marrow derived nuclear cells in the ONFH group at G0/G1 phase was significant increased (p<0.05) and S and G2/M significantly decreased after week 7. The rate of apoptotic cells in the ONFH group was higher than for the control group at week 7; however, the difference was not significant ( Figure 5 ).