The present study was approved by the Animal Experiment Committee of Xi'an Jiaotong University (Xi'an, China). The principles of laboratory animal care were followed, all procedures were conducted in accordance with the guidelines established by the National Institutes of Health and every effort was made to minimize the suffering of the animals. A total of 30 healthy, adult, male New Zealand white rabbits weighing 2.0–2.5 kg (mean, 2.25±0.15 kg) were supplied by the Center of Experimental Animals of Xi'an Jiaotong University School of Medicine. All rabbits were housed in standard cages with food and water ad libitum under a natural day/night cycle.

Hydrogen was dissolved in physiological saline for 6 h under a pressure of 0.4 MPa to a supersaturated level. The obtained hydrogen-rich saline was sterilized by γ-radiation and stored under atmospheric pressure at 4°C in an aluminum bag with no dead volume. The saline was freshly prepared every week to ensure a concentration of >0.6 mmol/l. Gas chromatography was used to confirm the content of hydrogen in the saline (5).

Rabbit thrombomodulin (TM) and vascular endothelial growth factor (VEGF) ELISA kits were purchased from Cusabio Biotech Co., Ltd. (Wuhan, China). Rabbit glutathione (GSH) and rabbit lipid peroxide (LPO) ELISA kits were obtained from Shanghai Bohua Biological Technology Co., Ltd. (Shanghai, China). Anti-CD34 polyclonal antibody (bs-0646R) was purchased from Boosen Biological Technology Co., Ltd. (Beijing, China). A 3,3′-diaminobenzidine (DAB) staining kit was obtained from Sequoia Jinqiao Biological Technology Co., Ltd (Beijing, China). Lipopolysaccharide (LPS) and prednisolone were purchased from Sigma-Aldrich (St Louis, MO, USA).

The animal model of steroid-associated osteonecrosis was prepared as previously described (15,16). Briefly, after 1 week of acclimation, all rabbits received one intravenous injection of LPS (10 µg/kg). After 24 h, an intragluteal injection of prednisolone was performed at a dosage of 20 mg/kg, once per day, for 3 days.

The animals were randomly divided into two groups (n=15 per group). Compared with the intravenous and local injections of hydrogen-rich saline, the intraperitoneal injection is more convenient for long-term treatment and was therefore selected in the present study. Rabbits in the hydrogen-rich saline group were treated with an intraperitoneal injection of hydrogen-rich saline at a dosage of 5 mg/kg/day, once per day, for 14 consecutive days, while those in the placebo group received normal saline.

Tests for GSH and LPO were performed immediately prior to the LPS injection and 3, 5, 7 and 14 days after saline treatment. At these time-points, blood samples (2 ml in each group) were collected from the auricular arteries of the rabbits under anesthesia, and plasma was obtained by transferring the blood sample into a tube containing 3.8% sodium citrate anticoagulant (0.2 ml) for centrifugation at 1,500 × g for 10 min at 4°C. The plasma was then stored at −70°C for the quantitative determination of GSH and LPO levels using ELISA. Measurements of VEGF and TM were performed prior to model establishment, as well as 2, 4 and 6 weeks after saline treatment. Plasma samples were obtained and used for the quantitative determination of VEGF and TM using the aforementioned method.

At 2, 4 and 6 weeks after saline treatment, 5 randomly selected rabbits from each group were sacrificed by an intravenous injection of air and the femoral heads were harvested. The femoral heads were fixed in neutral formaldehyde, decalcified, dehydrated with an alcohol series and embedded in paraffin. The samples were the cut into 5-µm sections along the coronal plane for the subsequent pathological staining using hematoxylin and eosin (H&E) and immunohistochemistry.

In order to highlight the microvessels, the sections were stained by immunohistochemistry for the detection of the angiogenic marker CD34 in endothelial cells. The deparaffinized and hydrated sections were washed with phosphate buffer solution three times, for 5 min each time, and were then treated with 3% hydrogen peroxide for 20 min to block endogenous peroxidase activity. The slides were immersed in 0.01 mol/l citrate buffer (pH 6.0), heated in a microwave at a constant temperature of 92–98°C for 10 min and then rinsed with phosphate buffer solution three times. After 1 h of incubation with 10% goat serum, the primary antibody was added (1:100) and the sections were incubated at room temperature for another hour, followed by overnight incubation at 4°C. Following three washes in phosphate buffer solution, the sections were incubated with biotinylated goat anti-rabbit IgG secondary antibody (1:200; bs-0295G-Bio; Boosen Biological Technology Co., Ltd., Beijing, China) for 2 h at room temperature, rinsed and placed in avidin-peroxidase conjugate solution for 2 h. The horseradish peroxidase chromogenic substrate 0.05% DAB was added for visualization. The sections were then counterstained with hematoxylin, dehydrated and mounted. Appropriate sections were selected to be used as positive and negative controls.

Microvessel counting was performed using the method previously reported by Weidner et al (17,18). Briefly, areas of the most intense CD34 antigen staining were identified by light microscopy at low power. The counting of the microvessels per field was performed using a higher-power microscope (magnification, ×200) manually from 5 random fields. The results are expressed as the average number of microvessels identified within the 5 fields. Any endothelial cell or cell cluster positive for CD34 antigen and clearly separate from an adjacent cluster was considered a single, countable microvessel.

All statistical analyses were performed using SPSS 16.0 software (SPSS Inc., Chicago, IL, USA). Data are expressed as the mean ± standard deviation. The comparison between the placebo and hydrogen-rich saline groups was performed using the Student's t-test. A one-way analysis of variance was used to compare matched data at multiple time-points. A two-tailed P<0.05 was considered to indicate a statistically significant difference.

There were no mortalities at any time during the experiment, and no side effects of the hydrogen-rich saline on the overall well-being of the rabbits were observed in either group.

The concentrations of VEGF and TM in the plasma decreased significantly following the hydrogen-rich saline treatment (Fig. 1). The differences between the hydrogen-rich saline and placebo groups were significant (P<0.05).

The activity of GSH and LPO was determined as an indicator of oxidative stress. As shown in Fig. 2, the GSH concentration in the plasma was significantly increased in the hydrogen-rich saline group at days 3 and 5 compared with that in the placebo group, and the LPO concentration was significantly decreased at day 5. No significant differences between the groups were observed on days 7 and 14.

Compared with baseline levels (immediately prior to the LPS injection), significant differences were observed in the GSH level in both the placebo and hydrogen-rich saline groups between days 3 and 7. With regard to LPO, a significant increase was only identified on day 5 in the placebo group. These differences gradually diminished with the passage of time.

The degree of osteonecrosis in the femoral head was assessed using H&E staining. The trabecular bone, lacunae, bone marrow and fat tissue were visualized using light microscopy (Fig. 3). Histological examination demonstrated that the condition of the femoral head in the placebo group became gradually aggravated as the time passed. A diffuse presence of empty lacunae or pyknotic nuclei of osteocytes was observed in the trabeculae at week 2, and thin trabeculae with a disordered texture and hypertrophic fat cells were noted. At week 4, thinner bone trabeculae were observed, with breakage of a part of the bone trabeculae. A number of lacunae were empty, and abundant adipose cells were found in the marrow cavity. At week 6, there was sparser trabecular bone with empty lacunae compared with week 4; however, the empty bone lacunae had clearly increased. The bone marrow had been partly replaced by fat tissue. There was no clear indication of new bone formation.

Histological examination revealed improvement in the trabecular bone, lacunae, bone marrow and fat tissue following treatment with hydrogen-rich saline. In the hydrogen-rich saline group, the bone trabeculae were thinner and no diffuse, empty lacunae were observed. The number of adipose cells had almost returned to normal at week 2. Some cells exhibited vacuolization or pyknosis at week 4, and empty bone lacunae were occasionally found. At the end of the experiment (week 6) a number of empty lacunae and osteocytes with pyknotic nuclei were found within the trabeculae. There were no signs of an inflammatory response (inflammatory cells).

Microvascular density was evaluated in order to ascertain vascular changes. All microvessels were highlighted using immunohistochemical staining. The results revealed a significant increase in the microvascular density in the hydrogen-rich saline group compared with that in the placebo group (Fig. 4) (P<0.05).