Micro arc-oxidized cylindrical pins (n=60; diameter 2.0 mm, length 6.0 mm, weight 0.500 g) made of AZ31 material were used (Trauson Medical Instrument Co., Ltd., Changzhou, China). AZ31 is a fast-degrading alloy of magnesium with 3 wt.% Al, 1 wt.% Zn and 0.15 wt.% Mn.

All animal experiments were conducted following ethical guidelines by Ethics Committee of Chinese PLA General Hospital (Beijing, China), obtained international standard authentication-SIDCER (54) and were authorized by the Institutes for Food and Drug Control of China and KEYU Animal Experiment Center.

A total of 60 male New Zealand white rabbits (body weight 2.2–2.5 kg, 3 months old) were purchased from KEYU Animal Experiment Center (Beijing, China; accreditation number SCXK(Jing) 2012-0004, certification no. 11400800001109) for the present study. Experiments were performed under standard conditions throughout the study (temperature, 23±2°C; relative humidity, 60±10%; with access to a 12-h light/dark cycle). Rabbits had been ensured adequate food and water. Each rabbit had an AZ31 pin implanted into its right femoral condyle. Rabbits were randomly divided into six groups (n=10), and were sacrificed after 1, 4, 12, 24, 36 or 48 weeks (1 group per time-point). Each pin was weighed prior to implantation and following sacrifice. Micro-CT, 3D reconstruction and histological examinations were performed following sacrifice.

Rabbits were fasted for 12 h prior to surgery and then anesthetized with an intraperitoneal injection of 3% sodium pentobarbital (Sigma-Aldrich; Merck KGaA; Darmstadt, Germany) with a dosage of 24 mg/kg. Surgery was performed under sterile conditions. First, full-thickness lesions were created with 2.0 mm Kirschner pins (Beijing Fule Science & Technology Development Co., Ltd., Beijing, China). AZ31 alloy cylindrical pins were implanted into the defects in the right femoral condyles. All pins were γ-ray-sterilized with 29 kGy of ⁶⁰Co radiation prior to surgery. All rabbits received an intramuscular anti-inflammatory injection with penicillin (Sigma-Aldrich; Merck KGaA) with a dosage of 400,000 U/day following surgery and were housed individually. Postoperatively, the rabbits were allowed to move freely in their cages without external support and with unrestricted weight bearing. Daily clinical observations were made throughout the study period.

Prior to implantation, the AZ31 pins were weighed on an electronic balance (accuracy, ±0.001 g). At 1, 4, 12, 24, 36 or 48 weeks, pins were removed from the femoral condyle and dry machined with clean tools. After machining, the pins were cleaned with pure ethanol in an ultrasonic bath and dried in warm air (23). The pins were weighed again and the difference between the pre- and post- implantation weights was calculated. To explore the difference between weight and Micro-CT methods, the pin weight fraction was calculated using the following formula: Weight loss of implanted pins/weight before implantation. Subsequently the pin weight fraction was compared with the pin volume fraction.

Micro-CT is an emerging technology that permits non-invasive, tissue-preserving imaging and quantitative morphometry of bone structure in three dimensions (55–56). Scans were performed with an RS-9 micro-CT (GE Healthcare, Chicago, IL, USA) to assess the condyle before it was placed in 25% formic acid solution for decalcification. The micro-CT system was operated at 80 kV tube voltage and 450 µA tube electric current, with a scan resolution of 45 µm and exposure time of 400 msec. Images were reconstructed and the pin volume, bone volume, BMD, tissue mineral density (TMD), BMC, BVF, BV/TV, BS/BV, Tb.Th, Tb.N, and Tb.Sp were analyzed using the built-in software (Version MicroView Advanced Bone Analysis Application 2.2; GE Healthcare) of the micro-CT equipment at each time point. A cylinder of the same size was selected from the corresponding region around the pins of the femoral condyle (Fig. 1A-F). A cylindrical ROI was set at each time point, matching the size of the magnesium alloy pins, 2.0 mm in diameter and 6.0 mm in length, ensuring that the ROI and pin fully overlapped (Fig. 1C). The threshold was set to 850 and the ROI was highlighted using the built-in software (Fig. 1D). The magnesium alloy was considered as bone tissue to assess degradation of the implant. Similar to the bone volume fraction, the Mg volume fraction was applied as: Degraded alloy pin volume/pin volume before implantation. The BMD and TMD reflected the Mg cylinder mineral and CT image densities, respectively. However, unlike natural bone, the pins had no trabecular, and thus the trabecular meant the thickness of the pin was calculated from the radius of its cross-section. To investigate new bone formation, the previous ROI was replaced with a larger ROI (diameter 2.5 mm, length 6.5 mm) in the same shape and position (Fig. 1E), from which new bone formation and the stimulatory effects on the growth of new tissue were observed. The threshold was then set at 1,000 and the ROI was highlighted with the built-in software (Fig. 1F). The maximum inner diameter of the new bone loop in the cross-section of the magnesium pin was selected, and the diameter of the pin was measured with the built-in software at the same position. Magnesium degradation and bone ingrowth were subsequently investigated.

The femoral condyles were excised and fixed in a 10% buffered neutral formalin solution for 1 week at room temperature at weeks 4, 24 and 48. Following decalcification, the samples were cut in half longitudinally and embedded in paraffin wax. Central sections (5 µm thick) were cut from the femoral condyle with a Leitz 1512 microtome (Leica Microsystems GmbH, Wetzlar, Germany) and stained at room temperature for 10 min with hematoxylin and eosin (H&E) for histological examination. Microscopic images were captured using a light microscope (dotSlide Virtual Slide System; Olympus Corporation, Tokyo, Japan) at ×40, ×100 and ×200 magnification.

Statistical analyses were performed using SPSS software (version 22.0; IBM Corp., Armonk, NY USA). Experimental values were expressed as the mean ± standard deviation and were analyzed using an unpaired Student's t-test to determine differences between the pre-implantation value and the post-implantation values at each time point. P<0.05 was considered to indicate a statistically significant difference.

According to the weight change of pins at each time point (Fig. 2), there was almost no weight loss of the implants within the first week (−0.007 g). However, by week 12, significant weight loss of the pins was observed compared with their pre-implantation weights (−0.017 g; P<0.05; Fig. 2). During weeks 12–24, the rate of weight loss was markedly increased when compared with weeks 1–4 and 4–12. After week 24, the rate of weight loss gradually decreased. This may be due to the production of new bone during corrosion of the micro arc-oxidized surface at the later time points.

The micro-CT images in Fig. 3 provide examples of the degradation process of an AZ31 implant at each time point. In the first week of the first month, little corrosion was observed, and the boundary of the pin was continuous and smooth (Fig. 3A). By week 4, the surface of the pin appeared blurred (Fig. 3B), indicating that the rate of degradation had increased compared with week 1. Corrosion pitting became evident at week 12 (Fig. 3C), and during weeks 12–24 the surfaces of the pins were surrounded by new bone tissue (Fig. 3C and D). These results suggest that degradation of the implants accelerated between weeks 4–24 and was markedly faster than that in the first 4 weeks. From week 36, corrosion pitting became more obvious, the boundary of the pin was inconclusive and the corners of the pins became indistinct (Fig. 3E and F). The images at this point suggested that degradation was still proceeding and the area of corrosion pitting was expanding. The acceleration of the degradation rate was observed at the preexisting corrosion points, and there did not appear to be corrosion at new positions on the surface of the pins. As indicated in Fig. 3F, corrosion pitting was evident on all surfaces and degradation was clear by week 48. However, the micro-CT images also identified a small amount of hydrogen gas around the pins at weeks 4, 12 and 48 only (Fig. 3B, C and F), however, no apparent trend was observed.

Similar to the micro-CT images, the data exported from the built-in software of the micro-CT equipment provided precise results (Fig. 4). The pin volume fraction was ≥99.0% during the first month of the study, indicating that degradation was slow and minor, and remained above 96.5% until week 12, after which the volume fractions were significantly lower when compared with the pre-implantation value (P<0.05; Fig. 4A). From week 12 to 24, the pin volume fraction decreased markedly faster than in the first 3 months, and reached 91.2% by week 24. During this period, the micro arc-oxidized surface was being degraded and new bone tissue had not yet been sufficiently produced (Fig. 3A-D). After week 24, the rate of decrease slowed, and during the final 24 weeks, pin volume fraction only decreased by 2.25% (Fig. 4A). This may have been due to enclosure of the residual pin by new bone tissue, thus reducing contact between the metal and tissue fluid.

The AZ31 magnesium alloy pin was regarded as bone to analyze the BMD and TMD, which reflected the Mg cylinder mineral and CT image densities, respectively. The Mg cylinder mineral density continuously decreased throughout the study period, degrading most rapidly between weeks 12 and 24. From week 12, the Mg cylinder material densities were significantly decreased compared with the pre-implantation value (P<0.05; Fig. 4B), and by week 48 it had reached 403.1424 mg/cc. By contrast, no significant change was observed in the pin CT image density, which started at 683.6439 (week 1) and ended at 644.9468 mg/cc (week 48; Fig. 4B). These results indicate that Mg content decreased as the implant degraded, while the density of the material underwent little change.

During degradation, the volume of the pin decreased, most notably during weeks 12–24 (Fig. 4A). Furthermore, with corrosion pitting, the surface of the pins became rough (Fig. 3C-F), which may increase the superficial area. Thus, the ratio between the surface area and volume of the pins increased throughout the study period, and were significantly higher compared with the pre-implantation value from week 24 (P<0.05; Fig. 4C). However, the surface area of the pins was preserved during the first 4 weeks (Fig. 4C). Unlike natural bone, the pins had no trabecular, and thus the trabecular mean thickness of the pin was calculated from the radius of its cross-section, and theoretically should be 1.0 mm. The computer creates a theoretical line that automatically runs through the longitudinal axis of the AZ31 pin, and the radius of its cross-section represents the pin's mean thickness. During the period investigated, degradation of the pins and increased corrosion pitting on the surface resulted in a decrease in the pin radius. The thickness of the pre-implantation pin was 1.0 mm and remained above 0.925 mm up to week 12, from which point it was significantly decreased compared with the pre-implantation value (P<0.05; Fig. 4D). During weeks 24–36, the thickness decreased more slowly than in 1–24 weeks, and during weeks 36–48, a slight increase in pin thickness was observed. This may have been due to the stimulatory effects of magnesium on the growth of new bone tissue at the metal corrosion loci.

To analyze pin degradation, the changes in pin weight fraction and volume fraction over the 48 weeks was plotted (Fig. 5). The results of two methods used (weighing and Micro-CT) to calculate amount of remaining pin were similar: both the weight and volume of the pins decreased over the 48 weeks, and during weeks 12–24, the decrease of the weight and volume of the pin was markedly fast. However, the data were not wholly consistent: at each time point, the quantity of magnesium corrosion by weight was significantly greater than by volume (P<0.05; Fig. 5).

A new, larger ROI was subsequently selected in the same shape and position as the original ROI (Fig. 1E), from which new bone formation and the stimulatory effects on bone tissue growth were observed. The threshold was set at 1,000, and the new ROI was highlighted with the built-in software (Fig. 1F). Before week 12, the majority of new bone formation occurred longitudinally on the surface of the metal, and the new bone appeared fragmentary and small (Fig. 6A and B). From week 4, a small amount of new bone began to grow transversely, and at the junction of the longitudinal and transverse profiles (Fig. 6B). Furthermore, the majority of new bone growth started at the bottom of the pin, which was in contact with the host cortical bone (Fig. 6B and C). From week 12, more new bone was formed, and by week 48, the pin was almost surrounded (Fig. 6C-E). The new bone growth was progressive and increased in all directions.

The data exported from the built-in software of the micro-CT equipment provided further precise results (Fig. 7). As indicated in Fig. 7A, from week 12, BVF values were significantly increased compared with the pre-implantation value (P<0.05). In addition, it was observed that osteogenesis increased slowly before week 24, while from week 24 to 36, the rate of new bone formation markedly increased (Fig. 7A). This may have been due to a faster corrosion rate and the stimulatory effects of released magnesium on the growth of new bone tissue. From week 36 to 48, the increase in BVF gradually slowed (Fig. 7A). This may be explained, to some extent, by the new bone preventing the magnesium alloy from degrading, to the disadvantage of further new bone formation. As new bone was produced, the superficial area of the ROI increased, and the ratio between the superficial area and volume of new bone (BS/BV) decreased, with significantly lower values of BS/BV observed from week 24 (P<0.05; Fig. 7B). This indicates that the new bone was gradually transforming from cribrate to compact. Fig. 7C-E illustrates the thickness, separation, and number of trabecular in the new bone over time. As the pin degraded, trabecular thickness and number significantly increased and separation significantly decreased from week 12 onwards (P<0.05; Fig. 7C-E). Concurrently to these changes, the TMD of new bone surrounding the pin significantly increased (Fig. 7F; P<0.05), indicating that the number of bones and new bone density increased as magnesium degraded.

In the cross-section of the magnesium pin, the maximum inner diameter of the new bone loop (Fig. 8A) and the diameter of the pin in the same position (Fig. 8B) were measured. During degradation, the diameter of the magnesium pin decreased. Theoretically, a firm attachment between the pin and new bone would cause the maximum inner diameter of the new bone loop to diminish, possibly faster than the diameter of the pin. As depicted in Fig. 9, the diameter of the magnesium pin decreased with degradation, from 1.98 mm in week 1 to 1.92 mm in week 48; however, no trend was observed in the maximum inner diameter of the new bone loop (max: 2.34 mm, week 36; min: 2.08 mm, week 1), instead of the expected downtrend. The two curves indicated that bridges had not been not adequately created by bone formation between the pin and surrounding tissues, although some bone and tissue contacts had appeared. It is probable that the biomaterials were not firmly attached to the surrounding tissues due to inadequate holding forces. This suggests that the magnesium alloy was not capable of creating sufficient bridges between the bones and biomaterials when there were preexisting gaps.

To investigate the association between material degradation and novel bone formation, the changes in pin volume fraction and bone volume fraction with time were plotted (Fig. 10). During the experiment, both rates were slow, and from week 1 to week 48, the volume of the magnesium alloy at each time point decreased by 10.58% (from 99.23 to 88.65%), whereas the volume of novel bone had increased by 21.54% (from 6.39 to 27.93%). By the end of the study, there was no intersection of the two curves (Fig. 10).

H&E staining was performed at weeks 4, 24 and 48. There were no apparent inflammatory cells surrounding the implants; however, chondrocytes and osteocytes were observed. No changes in the morphology of these cells between samples at 4, 24 and 48 weeks was observed (Fig. 11). Thus, there were no apparent inflammatory response around the implants. These findings suggested that the AZ31 magnesium alloys may be safe in human body.