ARTICLE IN PRESS
Biomaterials 26 (2005) 1043–1052
In vivo tissue response to resorbable silica xerogels as controlled-release materials Shula Radina,*, Gehan El-Bassyounia, Edward J. Vresilovicb, Evert Schepersc, Paul Ducheynea,b a
Center for Biomaterials and Tissue Engineering, Department of Bioengineering, University of Pennsylvania, 3320 Smith Walk, Philadelphia, PA 19104, USA b Center for Biomaterials and Tissue Engineering, Department of Orthopaedic Surgery, University of Pennsylvania, Philadelphia, PA 19104, USA c Department of Dentistry, Catholic University of Leuven, Leuven, Belgium Received 19 November 2003; accepted 6 April 2004 Available online 28 May 2004
Abstract Biodegradable, controlled-release carrier materials with non-toxic degradation products are valuable for local delivery of biologically active molecules. Previously, it was shown that room-temperature processed silica sol–gels (or xerogels) are porous, resorbable materials that can release molecules of various sizes in a controlled, time dependent manner. Previous in vitro studies also demonstrated benefits of silica xerogels as controlled-release materials for the treatment of bone infections. Herein the tissue and cell response to xerogels is documented using a subacute implantation procedure. The tissue response was correlated to composition, surface properties, resorption rate and incorporation of the antibiotic vancomycin. Ca- and P-free and Ca- and P-containing xerogels, with and without apatite (AP) surface, were used. Xerogels were implanted either as discs in a subcutaneous site, or as granules in the iliac crest of New Zealand white rabbits. The samples with surrounding tissue were retrieved after 2 and 4 weeks of implantation. Silica xerogels implanted either as discs subcutaneously or as granules in the iliac crest showed a favorable tissue response. The granules, either with or without Ca and P content, gradually resorbed over time. The resorption was accompanied by extensive trabecular bone growth and a minimal inflammatory response. Ca- and P-containing granules with an AP-surface layer showed a slower resorption rate and more extensive new bone growth than those without AP layer. Among AP-coated granules, those with incorporated vancomycin showed the most favorable tissue response. The present in vivo data together with prior in vitro data suggest that these xerogels have potential as controlled-release materials for the treatment of bone infections and as carrier materials for a variety of other applications. r 2004 Elsevier Ltd. All rights reserved. Keywords: Controlled-release materials; Silica xerogel; In vivo test; Bioresorption; Biocompatibility
1. Introduction Resorbable controlled-release materials are advantageous vis-a" -vis non-resorbable release materials as the need for their surgical removal is avoided [1–6]. There exist silica sol–gel materials that are resorbable [7,8]. They have been used for the encapsulation of enzymes, cells and living tissue [9–11]. Recently, room-tempera*Corresponding author. Tel.: +1-215-898-5140; fax: +1-215-5732071. E-mail address:
[email protected] (S. Radin). 0142-9612/$ - see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2004.04.004
ture synthesized silica sol–gels have been studied as controlled-release materials [12–18]. Upon hydrolysis of a silica precursor, biological molecules are added to the sol. Subsequent to condensation, aging and drying, the molecules are encapsulated in a solid sol–gel material. In vitro experiments, conducted in our laboratory, revealed that these sol–gel materials, also sometimes referred to as xerogels, could release low molecular weight drugs such as the antibiotic vancomycin (molecular weight: 2.8 kDa) [14,15]. Larger molecules, such as proteins and growth factors with a molecular weight of 20 kDa and larger could also be released in their functional form and
ARTICLE IN PRESS 1044
S. Radin et al. / Biomaterials 26 (2005) 1043–1052
in a controlled, time dependent manner [12,13]. Previously, we demonstrated that these xerogels were porous. Furthermore, we showed that their degradation in vitro depended on their pore characteristics and on their surface properties [7,8]. Specifically, the presence of an apatite (AP) surface film significantly reduced xerogel degradation. Altogether these in vitro studies document that silica xerogels are excellent resorbable, controlledrelease materials for a variety of applications, including the treatment of bone infections. To date, the in vivo behavior of silica xerogels as a controlled-release material has not yet been studied. Assessment of sintered sol–gel derived glasses (such as 58S and 77S glass) as bone graft materials has been reported [19], but these high temperature sintered glasses are fundamentally different from room-temperature processed xerogels. In this study, we determined the tissue response to various xerogels in the subacute implantation phase (up to 4 weeks of implantation). In addition, we set out to correlate the tissue response to xerogel composition, surface properties, resorption rate and incorporation of the antibiotic vancomycin. Ca- and P-free and Ca- and P-containing xerogels, with and without AP surface, were used. Although the initial results were presented earlier [20], this paper documents the results in full.
2. Materials and methods Xerogels were implanted either as discs in a subcutaneous site, or as granules in the iliac crest of New Zealand white rabbits. Xerogel composition, surface condition, implant shape and implantation site are summarized in Table 1. 2.1. Synthesis Xerogels were synthesized as described elsewhere [12– 15]. Briefly, the silica precursor tetramethylorthosilane (TMOS, Strem Chemicals, Newburyport, MA) was mixed with deionized (DI) water in a 1:10 molar ratio.
In all, 1 n HCl acid was used as a catalyst. When applicable, calcium chloride and triethyl phosphate (TEP, Strem Chemicals, Newburyport, MA) were added as calcium and phosphorus oxide precursors. Sol–gel compositions obtained upon drying were as follows: 100% SiO2 (S100), 90% SiO2–5% CaO–5% P2O5 (S90) and 85% SiO2–10% CaO–5% P2O5 (S85) (% in weight percent). When applicable, an aqueous solution of vancomycin (Eli Lilly & Co., Indianapolis, IN) was added to the sol prior to the addition of TEP and CaCl2. To form S90V composition, 20 mg vancomycin per milliliter of sol was added. The sols were cast in polystyrene vials. Discs, 8 mm diameter and 2 mm thick, resulting from gelling, aging and drying to constant weight, were crack-free. Some of the discs were crushed and sieved to produce granules in a 500–1000 mm size range. In order to form an AP-surface layer (denoted as AP in Table 1) Ca- and P-containing granules (S85, S90 and S90V) and discs (S90) were subjected to an immersion treatment. The solution consisted of Tris (Sigma, St. Louis, MO) buffered solution complemented with 2.5 mm Si and electrolytes in concentrations typical for plasma. We denote this solution TE-Si (for Tris, electrolyte and Si). For the treatment of vancomycincontaining S90V, 2 mg of vancomycin was added per milliliter of TE-Si solution (to minimize leaching of incorporated vancomycin during the treatment). Both granules and discs were immersed for 24 h. The granules were packed into small-bore syringes (4 mm diameter). Both granules and discs were sterilized by g-radiation at a dose of 30 kGy. 2.2. Material properties Synthesized xerogels were analyzed using FTIR spectroscopy (5DXC, Nicolet, Madison, WI) and gas (nitrogen) sorption analysis (Autosorb-1, Quantachrome, Boynton Beach, FL). The gas sorption analysis indicated that all xerogels were porous materials with a large surface area (SA) and nanosize pores: see Table 2. Ca- and P-free S100 xerogel
Table 1 Composition, surface treatment, shape and implantation site of xerogel implants Denomination
Oxide composition (wt%)
S100
100SiO2
S85-AP S90 S90-AP
85SiO2–10CaO–5P2O5 90SiO2–5CaO–5P2O5 90SiO2–5CaO–5P2O5
S90V S90V-AP
90SiO2–5CaO–5P2O5 90SiO2–5CaO–5P2O5
Shape +
Discs Granules+ Granules Discs Discs Granules Granules+ Granules+
Note: There were two phases of the implantation study as described in the text. All types of samples were used in phase I; only those marked with ‘‘+’’ were also used in phase II.
Implantation site Subcutaneous Bone Bone Subcutaneous Subcutaneous Bone Bone Bone
ARTICLE IN PRESS S. Radin et al. / Biomaterials 26 (2005) 1043–1052
was microporous (pore size o2 nm). Xerogels with Caand P-oxides were mesoporous (pore size between 2 and 50 nm). The physical properties of these xerogels are typical for acid-catalyzed silica xerogels [21]. FTIR analysis of xerogels suggested that they all were well polymerized and hydrolyzed. FTIR analysis also confirmed that after immersion treatment in TE-Si for 24 h, all Ca- and P-containing samples were coated with a layer of crystalline carbonate apatite (c-AP). The size of the granules was measured (in either the intact condition or sectioned after embedding in methylmethacrylate) using light microscopy. A semiautomatic image analysis system consisting of a highresolution color video camera (Toshiba 3-CCD) and Image-Pro Plus color/B&W image analysis software (Media Cybernetics, Silver Spring, MD) was used. Mean minimum and maximum cross-sectional lengths (Lmin and Lmax ) and mean SA of granules were determined. Dimensional values were determined on sections in order to serve as a comparative base to assess granule resorption on sections after implantation. The data for granules with and without the AP layer are shown in Table 3. The starting granules had an irregular and angular shape. The mean Lmax was approximately twice as large as the mean Lmin for all granules. It is logical that the dimensions measured on granule cross-sections were smaller than those measured on intact granules as cuts through granules pass at random, and not necessarily through the largest section. It was also found that the dimensions did not vary appreciably with granule composition or surface treatment. The thickness of the surface AP layer was determined on granule cross-sections at a magnification of 200 using a best-fit line mode. Five granules of each composition and at least 10 points per granule were used for the measurements. The thickness of the layer varied between 20 and 25 mm. Table 2 BET surface area (SA), pore volume (PV) and mean pore radius (PR) of as-synthesized xerogels Properties 2
SA (m /g) PV (103 cm3 g) PR (nm)
S100
S90
S85
842 410 1.0
304 285 1.9
270 370 2.8
1045
2.3. Implantation study The implantation proceeded in two phases: phase I was designed to screen the various xerogel compositions (Table 1) for the best tissue response; xerogels with an excellent tissue response in phase I were retained for phase II with a larger sample population. Two granule samples and four disc samples of each composition were used in phase I for each of two implantation durations, namely, 2 and 4 weeks. In phase II, three more samples were used for each experimental group. Phase II also included a sham implantation site as control. The materials used in phase I and those retained for phase II are listed in Table 1. Fourteen New Zealand white rabbits, weighing approximately 3 kg each, were used. Four and six animals were used per implantation time in phases I and II, respectively. The rabbits were housed according to NIH recommendations for the care and use of laboratory animals. Discs were implanted subcutaneously in the back. Granules were implanted in the iliac crest. Animals were anesthetized by a subcutaneous injection of 0.1 mg/kg atrophine and an intramuscular injection of 6 mg/kg xylazine, 50 mg/kg ketamine and 0.02 mg/kg acepromazine. Xerogel discs were implanted into subcutaneous pouches created from 2 cm long incisions made laterally on each side of the dorsum. The incisions were closed with resorbable sutures. Granules were implanted into cylindrical defects in the left and right iliac crest. A hand drill was used to create a defect of 5–6 mm in diameter and 2 mm in depth. After irrigation, xerogel granules were expelled from the syringes into the defects. The control defect sites were not filled. The wounds were closed in layers with resorbable sutures. All wounds were dressed using Vetbond tissue adhesive. Post-operative analgesia consisted of subcutaneous injection of Buprenorphine (0.03 mg/kg) immediately after surgery. When rabbits recovered after surgery, collars were placed around their neck. Penicillin was administered for 3 days postoperatively. 2.4. Histology and morphometry The animals were sacrificed at either 2 or 4 weeks. The subcutaneous implants with surrounding tissues and
Table 3 Mean surface area (SA, mm2) and minimum and maximum cross-sectional lengths (Lmin and Lmax ; mm) of granules measured on intact granules and cross-sections of embedded granules Material
S100 S90 S90-AP
Intact granules
Cross-sections
SA
Lmin
Lmax
SA
Lmin
Lmax
0.5670.13 0.5470.12 0.5370.11
0.5670.032 0.5270.034 0.4870.028
1.1270.1 1.0870.12 1.0670.04
0.3870.1 0.3370.09 0.3570.13
0.4070.08 0.4670.07 0.4270.07
0.9070.30 0.9470.26 0.9670.24
ARTICLE IN PRESS 1046
S. Radin et al. / Biomaterials 26 (2005) 1043–1052
bony blocks with the granules were immediately fixed in a solution consisting of one part formaldehyde, neutralized with 50 g CaCO3/l and two parts of 80% ethanol. The specimens were dehydrated in a series of graded alcohol. After being embedded in methylmethacrylate, the bony blocks were sectioned parallel to the axis of the cylindrical defects; subcutaneous implants were cut parallel to the disc plane. Five thin serial sections were cut per tissue block, then ground and polished to a thickness of 20–40 mm. The polished sections were stained with Giemsa or a combination of Stevenel’s blue with Von Gieson picro-fuchsin. The tissue and cell response were analyzed on thin stained sections. A total of 34 bone and 38 subcutaneous samples were used in both phases of the study. Three stained sections of each sample were used for the analysis. The inflammatory response was scored using the method of Royals et al. [6] on a scale from 0 to 4: 0 (absent); 1 (minimal)—presence of a few lymphocytes or macrophages, no presence of foreign-body giant cells (FBGC), eosinophils or neutrophils; 2 (mild)—several lymphocytes, macrophages, with a few FBGC and small foci of neutrophils; 3 (moderate)—large numbers of lymphocytes, macrophages and FBGC, with a number of neutrophils; 4 (severe)—tissue necrosis. Formations of new bone tissue within the defect and the granule size were measured light microscopically using the semiautomatic image analysis system. Parameters measured included the percentage of new bone tissue within the defect area, mean size of granules before and after implantation, and total area of granules within the defect area. The defect area available for bone growth was determined as a difference between the total defect area and the total granule area. The determination of the granule size included measurement of minimum and maximum cross-sectional lengths (Lmin and Lmax ) and apparent granule surface area (GSA). These parameters were determined for each granule within the defect area and then averaged for all granules present.
2.5. Statistical analysis Two-way analysis of variance (ANOVA) and paired Student’s t-test were used to test the difference between the experimental groups.
3. Results 3.1. Clinical observations The rabbits were followed up on a daily basis and no wound inflammation or other alarming clinical signs were observed. 3.2. Histological and histomorphometric evaluation— phase I The tissue response to implanted materials is illustrated by the micrographs of Figs. 1–3. The inflammatory response score is indicated in Table 3. 3.3. Subcutaneously implanted discs After 2 weeks of subcutaneous implantation, all the discs were encapsulated by a thin pseudo-synovial membrane of densely packed collagen fibers (Fig. 1). As shown on the micrographs (Fig. 1a and b) and as indicated by the score in Table 4, the S100 discs showed minimal inflammation after both 2 and 4 weeks of implantation. Ca- and P-containing S90 discs, with or without AP coating, also showed a minimal inflammation after 2 weeks; however, the number of inflammatory cells increased after 4 weeks (Table 4). For the APcoated discs, a greater number of inflammatory cells were observed in the areas, where separation of the surface AP film was observed. Evidence of significant disc resorption was not observed for any of the materials.
Fig. 1. Light micrographs of subcutaneous tissue response to S100 discs after 2 (a) and 4 (b) weeks of implantation. Original magnification: 100 . The discs were surrounded by a thin fibrous membrane, which contained few inflammatory cells after 2 weeks and very few after 4 weeks. Evidence of significant disc resorption was not observed after either implantation time.
ARTICLE IN PRESS S. Radin et al. / Biomaterials 26 (2005) 1043–1052
Fig. 2. Light micrograph of a cylindrical defect, created in the rabbit iliac crest, with S90V-AP granules after 2 weeks of implantation (phase I of the implantation study). Original magnification: 10 . Extensive trabecular bone growth was observed at the opening in the cortical bone, at the walls of the defects (where initially trabaculae are not present) and in the area adjacent to the original cortical bone. The trabecular growth along the walls of the defect resulted in walling off the defect from bone marrow.
Fig. 3. Higher magnification micrograph of S90V-AP granules and surrounding tissues after 2 weeks of implantation showing details of the tissue appearance. Bone trabeculae are covered with osteoid tissue and a row of active osteoblasts. The AP-surface layer (stained yellow, about 20 mm thick), which was formed on the granules before implantation, was not visibly changed after 2 weeks of implantation. There was the evidence of osteoconduction to the AP layer. Trabecular growth occurred in close vicinity to AP-coated granules of various compositions, however, osteoconduction was only observed for S90VAP granules. There were only few inflammatory cells at the granule surface and in the fibrous tissue between the granules.
1047
weeks of implantation. The lower magnification micrograph (Fig. 2) shows a cylindrical defect created in the iliac crest and S90V-AP granules implanted into the defect through the opening in the cortical bone. After 2 weeks of implantation, extensive trabecular bone growth was observed in the defect filled with granules. The growth occurred at the opening of the defect (where the periosteum was lifted), at the borders of the defect and in close vicinity to the cortical bone at the bottom. The trabecular growth resulted in walling off the defect from the bone marrow. The higher magnification micrograph (Fig. 3) shows that these trabeculae were covered with a layer of osteoid tissue and a row of active osteoblasts. New bone growth was extensive in close vicinity to APcoated granules of all compositions tested, however, osteoconduction was observed only for S90V-AP granules (Fig. 3). The higher magnification micrograph also demonstrates that the AP-surface layer (stained yellow) remained adherent on most granules. Coating thickness did not differ significantly among compositions and varied from 20 to 22 mm. Cracking and separation of the coating was observed in some locations after 4 weeks of implantation. The percentage of new bone in the defect as a function of implantation time and composition is represented in Fig. 4. All AP-covered granules showed a comparable trabecular bone growth after either 2 or 4 weeks of implantation, i.e. the variations among the experimental groups were not statistically significant. The morphometric analysis of the granule size showed a decrease in size with implantation time for all experimental groups used in phase I study. All granules used in phase I (S85-AP, S90-AP and S90V-AP) showed a minimal inflammatory response at either 2 or 4 weeks of implantation as indicated by the score in Table 4. Also, in comparison to granules without vancomycin, those with vancomycin showed an even lower inflammatory score (compare S90V-AP and S90-AP). The fibrous tissue between the granules contained few inflammatory cells and normal marrow components. However, some phagocytosing cells were observed at the granule surface in the areas adjacent to some granules with detached AP coating. In such areas, the number of phagocytosing cells appeared smaller for vancomycin-containing S90V-AP granules. According to the analysis, S100 discs and S90V-AP graft granules showed the most favorable tissue response in subcutaneous and bone implantation sites, respectively. These materials were selected for phase II of the experiments.
3.4. Granules implanted in the iliac crest
3.5. Histological and histomorphometric evaluation— phase II
Micrographs of Fig. 2 (10 ) and 3 (80 ) illustrate the tissue response to AP-coated xerogel granules after 2
In this phase of the study, xerogel S100 was used both as discs and as granules. S90V granules were used either
ARTICLE IN PRESS S. Radin et al. / Biomaterials 26 (2005) 1043–1052
1048 Table 4 Inflammatory response score Phase I Material
Control S85-AP S90-AP S90V-AP S100 S90 S90V
Phase II
Bone (granules)
Subcutaneous (discs)
Bone (granules)
Subcutaneous (discs)
2 weeks
4 weeks
2 weeks
4weeks
2 weeks
4 weeks
2 weeks
— 0.5–1.5 0.5–1.5 0.5–1.0 — — —
— 1–1.5 1–1.5 0.5–1.5 — — —
— — 0.5–1.5 — 0.5–1.5 0.5 —
— — 1.0–2.0 — 0.5–1.0 1.5–2.0 —
0 — — 1–1.5 1.0 — 1.5–2.0
0 — — 1–1.5 1.5 — 2.0
— — — — 1.0 — —
4 weeks
1.5
Movement of samples was apparent. More phagocytosing cells in areas with detached AP coating.
3.6. Subcutaneously implanted discs
Ratio of new bone/defect area,% 40
Ratio,%
with or without the AP film. Implantation procedure and implantation time were the same as in phase I. Light micrographs in Fig. 5a–f illustrate the bone tissue response to control (a, b), S100 (c, d) and S90V (e, f) granules without the AP layer. The inflammatory response score is shown in Table 4. Percentage of new bone in the defect is given as a function of implantation time in Fig. 6. Change in granule size (mean SA, mean Lmin and Lmax ) as a function of implantation time is shown in Fig. 7.
35
2 wks
30
4 wks
25 20 15 10 5 0 S90V-AP
The tissue response to S100 discs in phase II was similar to that observed in phase I. That is, the discs produced a minimal inflammatory response either at 2 or 4 weeks of implantation (Table 4).
S90-AP
S85-AP
3.7. Granules implanted in the iliac crest
Fig. 4. Percentage of new bone in the defect as a function of implantation time and experimental groups (AP-coated granules of various compositions, phase I). Error bars represent standard deviation (n ¼ 2). The differences among the experimental groups were not statistically significant (p ¼ 0:12). The data indicate that the healing of the defect via extensive trabecular bone growth was comparable for all AP-coated granules tested.
At 2 weeks, the control defect is healing via trabecular bone growth (Fig. 5a). The area is well vascularized; normal bone marrow component is present between the trabeculae. At 4 weeks, the amount and the density of bone trabeculae are greater than those at 2 weeks (Fig. 5b). The defect entrance appears repaired, however, the new cortical layer at the entrance is not yet as dense as the old cortical bone. Inflammatory cells were absent in the control samples at both 2 and 4 weeks. The bone tissue response to AP-coated S90V granules was similar to that observed in phase I. There was a minimal inflammatory response (Table 4) and extensive trabecular bone ingrowth. In addition, osteoconduction was observed. As far as S100 and S90V granules without the AP film are concerned, the optical micrographs in Fig. 5c–f illustrate the bone tissue response to these materials after 2 and 4 weeks of implantation. As in the case of AP-coated granules, healing of the defects via trabecular bone ingrowth also occurs in the presence of these
uncoated granules. However, in comparison to APcoated granules (Figs. 2 and 3), the trabecular bone growth appeared less extensive, and there was no evidence of osteoconduction. S100 granules showed a minimal inflammatory response at 2 weeks with a slight increase in the number of cells at 4 weeks (Table 4). There was a noticeable difference in the appearance of the S100 and S90V graft granules within the defect. After both 2 and 4 weeks, S90V granules appeared more spherical than the S100 granules which retained their original angular shape. For all experimental groups used in phase II, the progression of healing of the defect via trabecular bone growth was quantified as the percentage of new bone in the defect. The data are presented as a function of implantation time in Fig. 6. The percentage of new bone in the controls and for the defects filled with AP-coated S90V granules at both 2 and 4 weeks was statistically not different. The percentage of bone measured with
ARTICLE IN PRESS S. Radin et al. / Biomaterials 26 (2005) 1043–1052
(a)
(b)
(c)
(d)
(e)
(f )
1049
Fig. 5. Light micrographs of rabbit bone tissue response to (a, b) control, (c, d) S100 and (e, f) S90V granules without AP-surface layer after implantation into the iliac crest defects: for 2 (a, c, e) and 4 (b, d, f) weeks (phase II of the experiment). Original magnification: 12 (a, e, f) and 10 (b–d). Progressing healing of the defect via trabecular bone growth was observed for both the control and the implant groups. Extensive trabecular bone growth occurred at the opening and the borders of the defects at 2 weeks of implantation. After 4 weeks, the amount and the density of the trabaculae increased for all the experimental groups.
S90V-AP granules in phase II was comparable to that determined for AP-coated granules of various compositions in phase I (Fig. 4). In comparison, sites with uncoated S90V granules showed a tendency for slower healing, that is, the percentage of new bone in the defect area was significantly (po0:01) smaller than that for S90V-AP granules at both 2 and 4 weeks. Morphometric measurements of the granule size as a function of implantation time were conducted for all experimental groups used in phase II (Fig. 7). All granules gradually decreased in size with implantation time. The number of granules within the defect also decreased with time. The decrease in size for S100 was not statistically significant from that for S90V-AP granules. However, the size reduction was significantly greater (po0:05) for S90V granules without the AP film.
These observations indicate that degradation of S90V granules was apparently faster than that of S100 granules. After 2 weeks, the rate of the size reduction slowed down for all granules.
4. Discussion Biodegradable, controlled-release carrier materials with non-toxic degradation products are very valuable for local delivery of biologically active molecules. In response to the need for such carriers, biodegradable polymers have been proposed [1–5,22,34,35]. However, it has been reported that the degradation of polymers, which is the mechanism producing the molecule release, can cause an inflammatory response which interferes
ARTICLE IN PRESS S. Radin et al. / Biomaterials 26 (2005) 1043–1052
1050
Granule size vs. implantation time S100
2
40 35 30 25 20 15 10 5 0
0.4
Surface area, mm
2 weeks 4 weeks
S90V
0.3
S90V-AP 0.2
0.1
0 0
cntr
S100
S90V
1
S90V-AP
Fig. 6. Percentage of new bone in the defect as a function of implantation time and experimental groups (phase II: control, S100 and S90V with and without the AP layer). Error bars represent standard deviation (n ¼ 9). The data indicate that the rates of new bone growth were not statistically different for the controls and the defects filled with AP-coated S90V granules at both 2 and 4 weeks. Both S100 and S90V granules without the AP film showed a tendency for slower healing, however, only the difference between S90V-AP (with HA-layer) and S90V granules without the layer was statistically significant (po0:01).
2
3
4
Time, weeks 0.25 0.2 L min, mm
Ratio, %
Ratio of new bone/defect area (%) vs. implantation time
0.15 0.1 S100 S90V
0.05
S90V-AP
0 0
1
2 3 Time, weeks
4
5
0.6 S100
0.5
L max, mm
with the intended therapy [6,22]. Ceramic materials such as plaster of Paris, calcium phosphates (CP), glassceramic cements and others, have also been studied as carrier materials [23–29]. These materials are biocompatible and bone-bioactive; however, the ‘‘burst’’ release profile they usually exhibit, typically is a disadvantage for controlled-release applications. The current in vivo study further builds on our previous in vitro reports [12– 15] to document the exciting properties of silica xerogels as biocompatible controlled-release materials. Previous in vitro studies showed that both the rates of resorption and the release rates of xerogels depended on sol–gel processing parameters and surface conditioning [7,8,15]. A steady, long-term release of low molecular weight drugs (antibiotics) from xerogels was also demonstrated; the release occurred in the absence of a ‘‘burst’’ effect. Moreover, loaded molecules were also totally recovered during elution and retained their activity [14]. These observations suggested that the process used to synthesize the xerogel/molecule composites did not lead to molecule denaturation. Thus, neither the synthesis itself, nor the chemical reactivity of the matrix is of concern. The present in vivo study demonstrated a favorable tissue response to silica xerogels, implanted either as discs subcutaneously or as granules in bony sites. The granules showed a gradual resorption accompanied by extensive trabecular bone growth and a minimal inflammatory response. The gradual resorption of all xerogel granules followed from the morphometric analyses (Fig. 7). The granules remained intact, though. In addition, there was no evidence of bulk material breakdown. The morphometric analysis also showed that the extent of granule
S90V S90V-AP
0.4 0.3 0.2 0.1 0 0
1
2 3 Time, weeks
4
5
Fig. 7. Mean SA and mean minimum and maximum cross-sectional lengths (Lmin and Lmax ) of S100 and S90V granules with and without the AP layer as a function of implantation time (phase II). Error bars represent standard deviation (n ¼ 9). Granule size before implantation was measured on sections of embedded granules (Table 3). A gradual decrease in the granule size with implantation time was observed for all experimental groups. However, after 2 weeks of implantation S90V granules without the AP layer showed a statistically greater (po0:05) reduction in size than the granules with the layer.
resorption varied with granule composition (S90V vs. S100) and surface conditioning (AP-coated vs. uncoated). In fact, at 2 weeks of implantation, the rate of reduction in size was significantly greater for uncoated S90V than for AP-coated S90V. This observation suggests that the AP-surface film that protected silica xerogels against resorption in vitro [7,8], also had a protective effect in vivo. Concerning the compositional effect, the rate of reduction in granule size was significantly less for S100 than that for S90V. This suggests that the addition of Ca- and P-oxides to silica
ARTICLE IN PRESS S. Radin et al. / Biomaterials 26 (2005) 1043–1052
sol–gels affects the degree of xerogel resorption. This effect could be related to the observed change in physical properties of xerogels associated with addition of the oxides (Table 2). Whereas gradual xerogel resorption occurs both in vitro and in vivo, the granule resorption in vivo is slower than in vitro. In fact, total dissolution of S100 and S90 granules of similar size occurred after 1 week of in vitro immersion in physiological solution with solution exchange [7,8]. In contrast, only a decrease in size, not a total resorption, was observed for these granules after implantation up to 4 weeks. The slower reactivity in vivo may be associated with the effect of surrounding tissues on diffusion and mass transport of dissolution products. As an illustration, our laboratory reported increased concentrations of Si in muscle tissues surrounding silica-based glass implants [30]. Such concentrations then reduce the dissolution rate from the solid phase. In addition to xerogel degradation, there was progressive healing of the defect via trabecular bone growth. Histological analysis showed active formation of new bone tissue in close vicinity to xerogel granules (Fig. 3). As indicated by histomorphometric analysis (Fig. 6), the rates of new bone growth in the defects filled with granules were comparable to those in control sites. The data also indicate that the most extensive bone growth occurred during the first 2 weeks of implantation. All observations suggest that xerogel degradation does not adversely affect the healing process and the formation of new bone. Although the extensive trabecular bone growth was observed for all granules, the bone growth was statistically greater for AP-coated granules than that for granules without the AP coating, at both 2 and 4 weeks of implantation (Fig. 6). Also, the evidence of osteoconduction was observed only for AP-coated granules. This suggests that the AP layer formed prior to implantation positively influenced new bone tissue formation. Gradual degradation of xerogel granules was accompanied by a very mild inflammatory response. The gradual decrease in size and the limited number of inflammatory cells at the surface of xerogel implants suggest that controlled surface resorption via physicochemical reactions of silica dissolution was important in the resorption process. Detachment of the AP coating, which was occasionally observed, was associated with an increased number of phagocytosing cells. As detachment of this coating mostly occurred at sharp corners of the granules, the coating attachment can be enhanced by using spherical particles. The effect of incorporated vancomycin on the tissue response to xerogels is also noteworthy. Our previous in vitro studies demonstrated a controlled, long-term release of vancomycin in a therapeutically unaltered form
1051
[14,15]. This in vivo study confirmed a favorable tissue response to vancomycin-containing xerogels. Moreover, among various AP-coated xerogel granules tested, those with vancomycin (S90V-AP) showed the most favorable tissue and cell response. In fact, only these granules showed evidence of osteoconduction. In addition, the inflammatory score of these granules was lower. The subcutaneously implanted S100 xerogel discs elicited an excellent tissue response. These discs were surrounded by a thin fibrous membrane with fibroblasts and few inflammatory cells at both 2 and 4 weeks of implantation. In general, the inflammatory response is a part of the normal healing process. It includes an acute phase (0–7 days), subacute phase (7–28 days) and a chronic phase (beyond 28 days) [34]. The current study covers the subacute phase. Implantation in subcutaneous tissue is commonly used to assess biocompatibility. Various materials, including well-known biocompatible materials such as biodegradable and bioerodible polymers [34,35], hydroxyapatite [32] and bioactive glass (BG) [33], showed encapsulation with fibrous membranes of various thickness after short-term subcutaneous implantation. The tissue response to the xerogels compares favorably to the response of other biocompatible materials. In fact, the response to biodegradable polymers and BG was one with fibroblasts and mild inflammation with some macrophages [33,35]. Biodegradable polymers typically show extensive resorption and bulk material breakdown during shortterm subcutaneous implantation [34,35]. In contrast, subcutaneous xerogel implants of this study showed slow resorption and remained intact for the entire implantation duration of 4 weeks. This resorption behavior of xerogel implants is similar to that of highly compatible, resorbable ceramics and glasses [32,33]. Biocompatibility of dense silica-based biomaterials such as BGs and glass-ceramics has been confirmed in numerous studies [30,31,33]. It is well known that silicon is present in living tissues. In animals, it is abundant in connective tissue, hairs, skin, tendons, muscle and bone. It has been shown that upon degradation of dense silicabased glass in vivo, silica degradation products (in the form of monosilicic acid Si(OH)4) diffuse into the local tissue around the implant, enter the bloodstream or lymph and are then excreted in the urine [30]. The current in vivo study demonstrated biocompatibility of highly porous sol–gel derived silica xerogels. The path of removal of the xerogel degradation products, i.e. Si compounds, is expected to be similar to that documented for dense silica-based glasses.
5. Conclusions In summary, silica xerogel granules gradually resorbed with time and the defects healed continuously via
ARTICLE IN PRESS 1052
S. Radin et al. / Biomaterials 26 (2005) 1043–1052
extensive trabecular bone growth. Ca- and P-containing granules with an AP-surface layer showed a slower resorption rate and more intensive new bone growth than those without the AP layer. This implantation study demonstrates a favorable tissue response to silica xerogels, implanted either subcutaneously or in a bone site. As such, this in vivo data together with the previously published in vitro data suggest that these xerogels have potential as controlledrelease materials for the treatment of bone infections and as carriers for controlled delivery of drugs and larger biologically active molecules.
References [1] Lewis DH. Controlled release of bioactive agents from lactide/ glycolide polymers. In: Chasin M, Langer R, editors. Biodegradable polymers as drug delivery systems. New York: Marcel Dekker; 1990. p. 1–41. [2] Chasin M. Biodegradable polymers for controlled drug delivery. In: Hollinger JO, editor. Biomedical applications of synthetic biodegradable polymers. Boca Raton, FL: CRC Press; 1995. p. 1–17. [3] Heller J. Use of polymers in controlled release of active agents. In: Robinson JR, Lee VHL, editors. Controlled drug delivery, fundamentals and applications. New York: Marcel Dekker; 1987. p. 180–210. [4] Leong KW, Brott BC, Langer R. Bioerodible polyanhydrides as drug-carrier matrices. I. Characterization, degradation, and release characteristics. J Biomed Mater Res 1985;19:941–55. [5] Laurencin CT, Koh HJ, Neenan TX, Allcock HR, Langer R. Controlled release using a new bioerodible polyphosphazene matrix system. J Biomed Mater Res 1987;21:1231–46. [6] Royals MA, Fujita SM, Yewey GL, Rodriguez J, Schultheiss PC, Dunn RL. Biocompatibility of a biodegradable in situ forming implant system in rhesus monkeys. J Biomed Mater Res 1999;45:231–9. [7] Falaize S, Radin S, Ducheyne P. In vitro behavior of silica-based xerogels intended as controlled release carriers. J Am Ceram Soc 1999;82:969–76. [8] Radin S, Falaize S, Ducheyne P. In vitro bioactivity and degradation behavior of silica xerogels intended as controlled release materials. Biomaterials 2002;23:3113–22. [9] Braun S, Rappoport S, Zusman R, Avnir D, Ottolenghi M. Biologically active sol–gel glasses: trapping of enzymes. Mater Lett 1990;10(12):1–5. [10] Pope EJA. Living ceramics. J Sol–Gel Sci Tech 1995;4:225–9. [11] Pope EJA, Braun KP, Peterson CM. Bioartificial organs 1: silica gel encapsulated pancreatic islets for the treatment of diabetes. J Sol–Gel Sci Tech 1997;8:635–9. [12] Nicoll SB, Radin S, Santos EM, Tuan RS, Ducheyne P. In vitro release kinetics of biologically active transforming growth factorb1 from novel porous glass carrier. Biomaterials 1997;18:853–9. [13] Santos EM, Radin S, Ducheyne P. Sol–gel derived carrier for the controlled release of proteins. Biomaterials 1999;20:1695–700. [14] Radin S, Ducheyne P, Kamplain T, Tan BH. Silica sol–gel for the controlled release of antibiotics. I. Synthesis, characterization, and in vitro release. J Biomed Mater Res 2001;57:313–20. [15] Aughenbaugh W, Radin S, Ducheyne P. Silica sol–gel for the controlled release of antibiotics. II. The effect of synthesis parameters on the in vitro release kinetics of vancomycin. J Biomed Mater Res 2001;57:321–6.
[16] Bottcher H, Slowik P, Suss W. Sol–gel carrier systems for controlled drug delivery. J Sol–Gel Sci Tech 1998;13:277–81. [17] Kortesuo P, Ahola M, Kangas M, Kangasniemi I, Yli-Urpo A, Kiesvaara J. In vitro evaluation of sol–gel processed spray dried silica gel microspheres as carrier in controlled drug delivery. Int J Pharm 2000;200:223–9. [18] Ahola M, Kortesuo P, Kangasniemi I, Kiesvaara J, Yli-Urpo A. Silica xerogel carrier material for controlled release of toremifene citrate. Int J Pharm 2000;195:219–27. [19] Hench LL. Bioactive materials: the potential for tissue regeneration. J Biomed Mater Res 1998;41:511–8. [20] Radin S, El-Bassyouni G, Vresilovic EJ, Schepers E, Ducheyne P. Tissue reactions to controlled release silica xerogels carriers. In: LeGeros RZ, LeGeros JP, editors. Bioceramics, vol. 11. New York: World Scientific; 1998. p. 529–32. [21] Brinker CJ, Scherer GW. Sol–gel science. The physics and chemistry of sol–gel processing. San Diego: Academic Press; 1990. [22] Gombotz WR, Pankey SC, Bouchard LS, Phan DH, Puolakkainen PA. Stimulation of bone healing by transforming growth factor-beta1 released from polymeric or ceramic implants. J Appl Biomater 1994;5:141–50. [23] Cornell CN, Tyndall D, Waller S, Lane JM, Brause BD. Treatment of experimental osteomyelitis with antibiotic-impregnated bone graft substitute. J Orthop Res 1993;11:619–26. [24] Mackey D, Varbet A, Debeaumont D. Antibiotic loaded plaster of paris pellets: an in vitro study of a possible method of local antibiotic therapy in bone infection. Clin Orthop 1982;167: 263–8. [25] Shinto Y, Uchida A, Korkusuz F, Araki N, Ono K. Calcium hydroxyapatite ceramic used as a delivery system for antibiotics. J Bone Surg (Br) 1992;74:600–4. [26] Yamamura K, Iwata H, Yotsuyanagi T. Synthesis of antibioticloaded hydroxyapatite beads and in vitro drug release testing. J Biomed Mater Res 1992;26:1053–64. [27] Otsuka M, Matsuda Y, Kokubo T, Yoshihara S, Nakamura T, Yamamura T. Drug release from a novel self-setting bioactive glass bone cement containing cephalexin and its physicochemical properties. J Biomed Mater Res 1995;29:33–8. [28] Kimakhe S, Bohic S, Larrose C, Reynaud A, Pilet P, Giumelli B, Heymann D, Daculsi G. Biological activities of sustained polymixin B release from calcium phosphate biomaterial prepared by dynamic compaction: an in vitro study. J Biomed Mater Res 1999;47:18–27. [29] Benoit MA, Mousset B, Bouillet R, Gillard J. Antibiotic-loaded plaster of Paris implants coated with poly lactide-co-glycolide as a controlled release delivery system for the treatment of bone infections. Int Orthop 1997;21(6):403–8. [30] Lai W, Garino J, Ducheyne P. Silicon excretion from bioactive glass implanted in rabbit bone. Biomaterials 2002;23:213–7. [31] Schepers E, de Clercq M, Ducheyne P, Kempeneers R. Bioactive glass particulate as a filler for bone lesions. J Oral Rehabil 1991;18:439–52. [32] Li D-J, Ohsaki K, Ii K, Cui P-C, Ye Q, Baba K, Wang Q-C, Tenshin S, Takano-Yamamoto T. Thickness of fibrous capsule after implantation of hydroxyapatite in subcutaneous tissue in rats. J Biomed Mater Res 1999;45:322–6. [33] Wilson J, Pigott GH. Toxicology and biocompatibility of bioglasses. J Biomed Mater Res 1981;15:805–17. [34] Laurencin C, Domb A, Morris C, Brown V, Chasin M, McConnel R, Lange N, Langer R. Poly(anhydride) administration in high doses in vivo: studies of biocompatibility and toxicology. J Biomed Mater Res 1990;24:1463–81. [35] Ibim SM, Uhrich KE, Bronson R, El-Amin SF, Langer RS, Laurencin CT. Poly(anhydride-co-imides): in vivo biocompatibility in a rat model. Biomaterials 1998;19:941–51.
