Local application of rhTGF-β2 enhances peri-implant bone volume and bone-implant contact in a rat model

Bone 37 (2005) 55 – 62 www.elsevier.com/locate/bone

Local application of rhTGF-h2 enhances peri-implant bone volume and bone-implant contact in a rat model Aladino De Ranieria, Amarjit S. Virdia,b, Shinji Kurodaa, Susan Shottc, Robert M. Levena, Nadim J. Hallabb, Dale R. Sumnera,b,* a

Department of Anatomy and Cell Biology, Rush Medical College, Rush University Medical Center, Chicago, IL 60612, USA b Department of Orthopedic Surgery, Rush Medical College, Rush University Medical Center, Chicago, IL 60612, USA c Department of Internal Medicine, Rush Medical College, Rush University Medical Center, Chicago, IL 60612, USA Received 19 November 2004; revised 3 March 2005; accepted 7 March 2005 Available online 24 May 2005

Abstract Orthopedic and dental implant fixation depends upon bone regeneration. Growth factors such as transforming growth factor-beta (TGF-h) have been shown to enhance bone repair and strengthen the mechanical connection between implant and host skeleton in canine models. To provide a platform for studying molecular mechanisms of growth factor stimulated bone regeneration and implant fixation, the present study examined peri-implant bone volume as a response to TGF-h treatment in a rodent model. The rat femoral ablation model in which an implant is placed in the medullary cavity of the femur was used to examine the dose response to TGF-h2 applied to the implant (0, 0.1, 1.0, or 10 Ag). The study included a total of 40 rats (10 per dose) examined at 28 days. Peri-implant bone volume and bone-implant contact were assessed through microcomputed tomography and implant fixation strength was determined by a mechanical pullout test. Treatment of the implant with 10 Ag TGF-h2 led to a 2-fold increase in bone volume ( P < 0.001) and a 1.5-fold increase in bone-implant contact ( P < 0.01) with a trend of increasing fixation strength (non-significant increase of 1.4-fold). TGF-h2 treatment with 10 Ag led to uniform peri-implant bone volume and bone-implant contact along the length of the implant, whereas the other groups had less bone at the mid-point compared to the proximal and distal aspects of the implant. About 50% of the variance in implant fixation strength was explained by a regression model involving both bone-implant contact and peri-implant bone volume. D 2005 Elsevier Inc. All rights reserved. Keywords: Implant fixation; TGF-h; Microcomputed tomography; Bone regeneration

Introduction Orthopedic and dental implants rely on bone regeneration to achieve mechanical fixation to the host skeleton, a requisite for clinical success. A variety of anabolic agents are now known to enhance bone regeneration. Recently, considerable efforts have been made to enhance and accelerate intramembranous bone regeneration, the particular form of bone repair important for implant fixation. * Corresponding author. Department of Anatomy and Cell Biology, Rush Medical College, Rush University Medical Center, 600 S Paulina, Room 507, Chicago, IL 60612, USA. Fax: +1 312 942 5744. E-mail address: [email protected] (D.R. Sumner). 8756-3282/$ - see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.bone.2005.03.011

Several growth factors, including members of the TGF-h superfamily, have demonstrated this ability in canine models [1 –4]. These studies have shown that local application of growth factor to the implant is an effective strategy to enhance peri-implant bone volume, bone-implant contact, or bone ingrowth as well as the mechanical force required to dislodge the implant from the host skeleton. There is a need for a small animal model to study enhanced bone regeneration and implant fixation so that molecular biology endpoints can eventually be more easily examined. In the present study we have adapted a rat model used originally to study hematopoiesis [5,6] and more recently used to study intramembranous bone formation [7– 10]. A number of groups are now using an adaptation of this

56

A. De Ranieri et al. / Bone 37 (2005) 55 – 62

model to examine issues related to implant fixation, with the main focus to date being on implant surface treatments [11– 15]. This is an appropriate model for studies of implant fixation because intramembranous bone formation is a basic requirement for ‘‘cementless’’ implant fixation. Although growth factors have not received much attention in the rat marrow ablation implant model, there is one recent study in which TGF-h doses up to 1 Ag were reported to have no effect [16]. The purpose of the present study was to determine if local application of TGF-h in the rat model led to increased peri-implant bone volume, bone-implant contact, and/or implant fixation strength. Accordingly, we performed an experiment to determine if doses of TGF-h2 ranging from 0.1 to 10 Ag/implant were effective in enhancing bone volume, bone-implant contact, or implant fixation strength at 28 days in the rat model.

sterile and endotoxin free. The implants were air dried and aseptically packaged. All handling of the sterile implants took place in a class 100 laminar flow hood. An in vitro release kinetic experiment was performed to determine the amount of TGF-h2 released from the implant over time. The study used one implant loaded with 1 Ag of growth factor and one implant loaded with 10 Ag of growth factor, as described above. The experimental procedures were similar to those previously described in an in vitro study with a canine implant [18]. Briefly, the release profile was created by incubating the implants in phosphate-buffered saline with 1% bovine serum albumin at 37-C on a rocker. The sample implants were placed into fresh aliquots of buffer at regular intervals (0, 3, 6, 12, 24, 48, 96, 120, and 168 h). An ELISA using the TGF-h-specific antibody, 1D11 (Genzyme Corporation, Framingham, MA), was performed to assess the temporal release. Rat model

Materials and methods Forty 6-month-old male Sprague – Dawley rats were implanted unilaterally with hydroxyapatite/tricalcium phosphate (HA/TCP)-coated titanium rods treated with buffer only as a control, 0.1 Ag, 1 Ag, or 10 Ag recombinant human TGF-h2 in an IACUC-approved study. Animals were killed 4 weeks post-surgery. For each group, seven implanted femurs were first analyzed by microcomputed tomography (ACT) to measure bone volume near the implant and boneimplant contact, followed by a mechanical pullout test to measure the strength of fixation of the implant. The remaining three specimens were embedded in plastic, sectioned with a diamond band saw (Exakt Model 300CP) into 1 mm thick slabs, ground to a nominal thickness of 100 Am (Buehler model Phoenix 4000, Lake Bluff, IL), and stained with toluidine blue and basic fuchsin, following methods previously described in detail [17]. Implants, TGF-b loading, and in vitro release Titanium implants, 22 mm long and 1.5 mm diameter (Grandis Metals, Lake Forest, CA), were coated with an HA/TCP surface by the plasma flame spray technique (Zimmer, Warsaw, IN). The HA/TCP layer was approximately 50 Am thick and has been determined to contain 80% hydroxyapatite, 15% tricalcium phosphate, and 5% uncharacterized calcium phosphates [1]. The target doses of recombinant human TGF-h2 (a gift of Genzyme Corporation, Framingham, MA) were applied by pipetting 5.88 Al volumes of diluted stock solutions uniformly onto the HA/ TCP surface of each implant. The stock solutions were formulated in a 30-mM Na citrate/3% mannitol, pH 2.5 buffer. Control implants were treated with 5.88 Al of buffer. The volume used for the TGF-h2 stock solution and the control buffer was small enough to allow even coating of the implant without saturation or dripping. The solutions were

Male Sprague –Dawley rats (400 –450 g, approximately 6 months old, Harlan, Indianapolis, IN) received unilateral femoral implants in the left femur in an IACUC approval study. The implantations were performed under anesthesia using ketamine (100 mg/kg ip) and xylazine (5 mg/kg ip), supplemented as necessary. Each hind limb was first shaved and then scrubbed with Betadine and alcohol. Using aseptic technique, a 1-cm incision medial to the patella was made to expose the knee joint. A 2-mm hole was drilled into the patellar groove with a Dremel drill bit to penetrate the subchondral cortical bone and gain access to the femoral medullary canal. The marrow cavity was disrupted by inserting a threaded hand drill proximally through the entire length of the diaphysis to approximately the level of the lesser trochanter. A guide implant was placed into the ablated cavity to ensure that the canal was an appropriate size to accommodate the definitive implant. The cavity was then flushed with 10 ml of sterile saline for removal of loose marrow contents. Following irrigation, an implant was placed into the canal. Bone wax was used to plug the distal end of the femur, and vicryl sutures protected by staples were used to seal the wound. At 28 days, the rats were euthanized in a carbon dioxide chamber. The femurs for ACT and mechanical testing were harvested, denuded of soft tissue, and frozen at 20-C. The femurs for histology were placed in 10% neutral buffered formalin. Microcomputed tomography (lCT) and determination of peri-implant bone volume and bone-implant contact ACT (Scanco 40, Scanco USA, Wayne, PA) scans were taken at seven evenly spaced slices along the length of the implant (Fig. 1). The scan resolution was 16 Am in plane with a slice thickness of 16 Am. At the proximal four sections, the slices were analyzed for bone volume per total volume (BV/TV) after masking out the implant and adjacent

A. De Ranieri et al. / Bone 37 (2005) 55 – 62

57

bone-implant contact was performed by one individual who was blinded to the treatment status and we have found a high intraobserver degree of reproducibility (r = 0.928, P < 0.001 between repeated measurements). Mechanical pullout test Following ACT scanning, a mechanical pullout test to measure the strength of fixation of the implant to the host bone was performed. To prepare the specimen, the proximal part of the femur was embedded in dental acrylic (Lang Dental, Wheeling, IL) and 2– 4 mm of the implant was exposed. The exposed implant was gripped and S hooks were placed at either end of the specimen to permit coaxial alignment of the implant with the direction of force. Pullout testing was conducted at a displacement rate of 0.25 mm/ min to failure with the force recorded in Newtons (Instron model 8871, Canton, MA). Because the part of the implant exposed for gripping varied, we report shear strength as calculated by dividing the force (N) at the point of failure by the surface area of the implant nominally in contact with tissue (implant circumference  implant length not exposed for gripping). One specimen in the 1.0 Ag was not included

Fig. 1. Contact radiograph of a rat femur with implant. The lines show the locations of the microcomputed tomography slices. Slices 1 – 4 were used for measurement of bone volume in the area between the implant and endocortical surface. All 7 slices were used for measurement of boneimplant contact. Scale bar, 5 mm.

16 voxels and the cortical bone to define the total volume (MatLab, MathWorks, Inc., Natick, MA). The adjacent 16 voxels were masked in addition to the implant because test images found metal-induced artifacts in these voxels, consistent with a recent report [19]. BV/TV was calculated by binarizing the region of interest at a threshold equivalent to ¨175 in the Scanco software. This area of the femur is normally devoid of bone so the assumption was made that bone found within the medullary cavity in sections 1 through 4 had formed since surgery. Bone-implant contact was measured at all seven levels by using a grid with test lines that radiated outward from the center of the implant and counting the number of test line-implant interface intersections positive for bone and dividing this sum by the total number of test line-grid intersections. The criteria for assigning an intersection positive for bone were conservative, requiring that the intersection have morphology typical of bone as well as the appropriate grey scale intensity. For bone-implant contact, the voxels immediately adjacent to the implant were not excluded as we assumed that boneimplant contact could be judged by eye. Assessment of

Fig. 2. Release kinetics for (A) an implant treated with 1.0 Ag TGF-h2 and (B) an implant treated with 10 Ag TGF-h2. The cumulative release is plotted as a function of time. Note the 10-fold difference in scale for the vertical axis in the two graphs.

58

A. De Ranieri et al. / Bone 37 (2005) 55 – 62

Statistical analyses The morphometric and mechanical data appeared to be normally distributed and were, therefore, analyzed by analysis of variance with Bonferonni corrected post hoc tests, repeated measures analyses of variance and stepwise multiple regression (SPSS version 11.0 for Windows, SPSS, Inc., Chicago, IL).

Results

Fig. 3. Microcomputed tomography images from slice 3 at 4 weeks. (A) Control implant treated with buffer only. (B) Implant treated with 0.1 Ag rhTGF-h2. (C) Implant treated with 1 Ag rhTGF-h2. (D) Implant treated with 10 Ag rhTGF-h2. Note the increased bone formation adjacent to the implant treated with the highest dose of the growth factor. Scale bar, 2 mm.

because of a technical problem during testing. To determine the baseline strength of fixation, implants were placed in 10 rat femurs ex vivo following the surgical protocol except that the site was not irrigated. These specimens were then subjected to the pullout test.

In both the 1-Ag and 10-Ag loaded implants, 25– 30% of the adsorbed TGF-h was released into the medium, predominately within the first 48 h (Fig. 2). There was only very slow release after 48 h. Inspection of the ACT images showed that implants treated with the 10-Ag dose of TGF-h2 typically had a zone of mineralized bone in the peri-implant region, but the lower dose and control groups had no apparent increase in bone (Fig. 3). Histology showed that there typically was a very thin layer of bone that had formed on the HA/TCP surface whether or not the implant had been treated with TGF-h2 (Figs. 4A and B). This layer was about 5 –10 Am thick. Along most aspects of the implant there were few apparent connections between this thin layer and the surrounding bone, except in the 10-Ag dose group. In addition, in the 10Ag dose group, there was often a rim of bone separated from the implant by a space 50 –200 Am wide that was often filled with apparent unmineralized osteoid (Figs. 4C and D).

Fig. 4. Photomicrographs of the bone-implant interface. (A) Control implant treated with buffer (scale bar, 300 Am); (B) enlarged region from panel A, showing thin layer of bone at interface (see arrow) (scale bar, 100 Am); (C) implant treated with 10 Ag TGF-h2, showing gap between newly formed bone and implant (see asterisk) (scale bar, 300 Am); (D) enlarged region from panel C, showing that the gap region is occupied by apparent osteoid (scale bar, 100 Am).

A. De Ranieri et al. / Bone 37 (2005) 55 – 62

The quantitative study of BV/TV showed dose was a significant factor ( P < 0.001 in the analysis of variance) with the 10-Ag group having significantly elevated BV/TV compared to each of the other groups ( P < 0.001) (Fig. 5A). Similarly, bone-implant contact varied as a function of dose ( P < 0.001 in the analysis of variance) with the 10-Ag group having significantly elevated bone-implant contact compared to each of the other groups ( P < 0.01) (Fig. 5B). The distribution of bone varied along the length of the implant in two basic patterns. In the control group, BV/TV

Fig. 5. (A) Bone volume per tissue volume (BV/TV) for the region between the implant and endocortical surface increased as a function of dose ( P < 0.001 in the one-way ANOVA); **group different than each other group ( P < 0.001). (B) Bone-implant contact increased as a function of dose ( P < 0.001 in the one-way ANOVA); *group different than each other group ( P < 0.01). (C) Strength of fixation of the implant tended to increase with dose, but the effect was not statistically significant. Note: for all three graphs, means and standard deviations are plotted.

59

(which was only measured at sections 1 through 4) was highest proximally followed by a drop off in sections 2 through 4 (Fig. 6A, top panel). A similar pattern was observed for BV/TV for all growth factor-treated groups except the 10-Ag group which showed a more uniform distribution of BV/TV in sections 1 through 4 (Fig. 6A, bottom panel). This observation was confirmed by the repeated measures analyses of variance which showed significant section location effects for BV/TV for the 0Ag, 0.1-Ag, and 1.0-Ag groups ( P < 0.001 in each group), but not for the 10-Ag group ( P = 0.456). Bone-implant contact (which was measured at all 7 sections examined by ACT) was highest proximally (section 1) and distally (section 7) in the control group with less contact toward the middle of the implant (Fig. 6B, top panel). The distribution of bone-implant contact was similar in the other treatment groups except the 10-Ag treatment group which had a more uniform distribution of bone-implant contact (Fig. 6B, bottom panel). Statistically, there was a significant within-subjects section effect for bone-implant contact for the 0-Ag, 0.1-Ag, and 1.0-Ag groups ( P < 0.001), but not for the 10-Ag group ( P = 0.097). The mechanical pullout results showed a non-significant dose-dependent trend of elevated strength (Fig. 5C; P = 0.480 in the analysis of variance). One of values for the 1.0-Ag group was very low compared to the other cases in this group (0.15 MPa vs. a range of 1.17 –2.32 for the other cases). If this case is eliminated as an outlier, then this group had the highest implant strength of fixation (1.87 T 0.45 vs. 1.69 T 0.61 MPa for the 10-Ag group and 1.22 T 0.42 MPa for the control group), although statistically there still was not a significant dose effect ( P = 0.132 in the analysis of variance). The baseline strength of fixation strength was 0 MPa in 9 of the 10 bones implanted ex vivo and was less than 0.02 MPa in the other bone. Implant fixation strength was dependent upon BV/TV and bone-implant contact. In the multiple regression model in which BV/TV at each of the four sections measured and bone-implant contact at each of the 7 sections measured were allowed to enter the equation, a model involving boneimplant contact at sections 3 and 7 and BV/TV at section 1 was found to account for 52% of the variance in implant fixation strength ( P < 0.001; Table 1). The strongest bivariate correlation with strength was for bone-implant contact at section 3 (r = 0.547, P = 0.002), implying that this variable alone accounted for nearly 30% of the variance in implant fixation strength.

Discussion The present study demonstrated that treatment of the implant with 10 Ag TGF-h2 resulted in enhanced periimplant bone volume and bone-implant contact in the rat model. Lower doses (1 Ag and 0.1 Ag) were not effective.

60

A. De Ranieri et al. / Bone 37 (2005) 55 – 62

Fig. 6. Distribution of (A) bone volume/total volume (BV/TV) and (B) bone-implant contact as a function of section location for each experimental dose. Note: for all graphs, means and standard deviations are plotted.

There was only a trend toward increased mechanical fixation of the implant, with a slight dose dependency. The morphologic findings are consistent with previous work in canine models in which either TGF-h1 or TGFh2 applied to the implant led to increased bone formation

[1 – 3,18,20,21]. Experiments on enhancement of periimplant bone formation (intramembranous repair) with TGF-h in other species are not common, with negative findings reported for sheep [22] and rats [16]. Positive results on fracture healing (endochondral repair) following release

A. De Ranieri et al. / Bone 37 (2005) 55 – 62 Table 1 Results from the step-wise multiple regression for predicting implant fixation strength Model (variables entered)

Adjusted r 2

Significance of the change in r 2

Bone-implant contact at section 3 Bone-implant contact at section 3 BV/TV at section 1 Bone-implant contact at section 3 BV/TV at section 1 Bone-implant contact at

0.270

0.004

0.004

0.434

0.010

0.001

0.515

0.038

<0.001

Overall significance

section 7

of TGF-h from a medullary implant have been reported in rats [23]. The dose range used in the sheep study [22] overlapped the effective doses used in the canine studies so it is not clear why no benefit was found in the sheep study. The maximum dose used in the rat study [16] was 1 Ag and, in consideration of the findings from the present study, may simply have been too low to elicit a morphologic or mechanical effect. A limitation of the rat model used in the present study is that the magnitude of the anabolic effect appears to be lower than found in canines for the mechanical endpoint of implant fixation strength. The fixation strength of the implant in the 10-Ag group was a non-significant 39% higher than the control group in the present experiment. In contrast, in the canine model, implant fixation strength can be increased by as much as 3-fold following growth factor treatment [3], consistent with 2- to 4-fold increases in BV/ TV within the gap in canine models [1,3,18]. Interestingly, BV/TV in the 10-Ag group in the present study was more than 2-fold greater than the control, a value consistent with findings for the gap region in canine models. We propose that differences in bone-implant geometry in the rat and canine models may account for the different mechanical responses in these two species. In the rat model, it is likely that some degree of mechanical fixation occurs proximally and distally regardless of treatment because of the proximity of the implant to the host bone in these locations, although the baseline mechanical fixation was nil. Nevertheless, the close proximity of the implant to the host bone may account for the finding of higher bone-implant contact at these sections in the present study and previous studies from our laboratory than at the midpoint of the implant [14,15]. Furthermore, in these two previous studies, only bone-implant contact from the most proximal and distal sections was predictive of implant fixation strength [14,15]. In contrast, in the present study in which bone regeneration was enhanced, bone-implant contact at section 3 (the midpoint of the implant) was the first variable to enter the regression model predicting implant fixation strength. In the canine model the greater absolute size of the bone allows for creation of a controlled defect, meaning that proximity to host bone is always low and that treatment-induced

61

increases in implant fixation strength are greater than in the rat model. Thus, the anatomy of the rat model, involving some degree of inherent proximal and distal implant proximity to host bone, limits to some extent the magnitude of increase achievable with mechanical endpoints. Another contributing factor limiting the sensitivity of the mechanical measurements may have been the presence of the HA/TCP coating on the implant (the carrier for the growth factor). Calcium phosphate coatings are known to enhance implant fixation [24,25] and the HA/TCP coating used in the present study was found to enhance mechanical fixation by 50% compared to non-HA/TCP treated implants in the rat model (unpublished data). The data from the present study may be useful in the design of future studies. For instance, for the 39% difference in implant fixation strength between the control group and 10 Ag treatment group to be significant, given the relatively high variance for this measurement, a sample size of 20 per group would be needed (assuming a = 0.05 and b = 0.80) [26]. Even if we make this calculation using the data from the 1.0-Ag group with the one outlier removed (i.e., where the difference between this group and the control group was 53%), a sample size of 9 per group would be needed to demonstrate an effect. Thus, should the goal be to show directly an increase in mechanical fixation strength, future experiments need to be designed with greater sample sizes. Clearly, the use of morphological data (in this case, BV/ TV, and bone-implant contact) as opposed to mechanical data permits discrimination between TGF-h treatment groups in the rat model with smaller sample sizes than required for the mechanical endpoint on implant fixation strength. It is possible that creating more connections between the regenerating bone and the host bone would lead to a more profound mechanical effect. However, given that approximately 50% of the variance in implant fixation strength was associated with variance in bone-implant contact and peri-implant BV/TV in the present study, it is likely that use of TGF-h did have a mechanical effect, but that it simply was not directly detectable. The in vitro growth factor release profile found in the present study was consistent with previous work in our laboratory in which canine implants were used [18] as the release occurred rapidly with essentially no sustained release beyond 48 h. In the rat model, there was approximately 30% cumulative release as opposed to approximately 15% cumulative release in the canine model [18]. If one assumes the released TGF-h is responsible for the biological activity and that the in vitro release profile is characteristic of the in vivo profile, then TGF-h’s effect may be primarily during the early phase. The finding that the transcription profiles for the inflammatory marker gene, cyclooxygenase-2, was altered at early time points following TGF-h treatment in this model is consistent with this interpretation [27]. Alternatively, it is possible that it is the growth factor retained on the implant that causes the

62

A. De Ranieri et al. / Bone 37 (2005) 55 – 62

enhancement of bone formation, implying that the anabolic stimulus was available over a more prolonged time frame. The cascade of morphological events following marrow ablation is well defined [5 –9]. In the rat femoral implant model, the basic regenerative cycle is similar to the marrow ablation model in that the surgical injury induces medullary bone formation within 14 days, followed by a period of bone remodeling during which much of the medullary bone is removed, although some bone is left in contact with the implant [11,14]. In the present study, a relatively large volume of peri-implant bone was still present at 28 days in the 10-Ag TGF-h group compared to the control group and to previous rat implant studies [11,14]. The likely mechanism of action can be inferred from a recent study in the same model which found that application of TGF-h2 to the implant led to up-regulation of gene expression for several markers of bone formation between days 7 and 14 (collagen type I, osteocalcin, alkaline phosphatase, osteonectin, osteopontin, and core binding factor 1) without late suppression of markers often associated with bone remodeling (tumor necrosis factor alpha and cyclooxygenase-2) [27]. Thus, it is likely that the 10-Ag TGF-h dose stimulated bone formation without altering bone resorption, resulting in prolonged persistence of peri-implant bone volume.

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

Acknowledgments Supported by the National Institutes of Health grants AR42862 and RR16631 and the Grainger Foundation. Chris Hendrick, Eileen Broderick, Amit Ailiani, and Susan Infanger provided technical assistance. Implant coatings were a gift of Zimmer (Warsaw, IN) and recombinant human TGF-h2 was a gift of Genzyme Corporation (Framingham, MA).

References [1] Sumner DR, Turner TM, Purchio AF, Gombotz WR, Urban RM, Galante JO. Enhancement of bone ingrowth by transforming growth factor beta. J Bone Joint Surg 1995;77-A:1135 – 47. [2] Lind M, Overgaard S, Ongpipattanakul B, Nguyen T, Bu¨nger C, Søballe K. Transforming growth factor-h1 stimulates bone ongrowth to weight-loaded tricalcium phosphate coated implants. J Bone Joint Surg 1996;78-B:377 – 82. [3] Lind M, Overgaard S, Søballe K, Nguyen T, Ongpipattanakul B, Bu¨nger C. Transforming growth factor-h1 enhances bone healing to unloaded tricalcium phosphate coated implants: an experimental study in dogs. J Orthop Res 1996;14:343 – 50. [4] Sumner DR, Turner TM, Urban RM, Turek T, Seeherman H, Wozney JM. Locally delivered rhBMP-2 enhances bone ingrowth and gap healing in a canine model. J Orthop Res 2004;22:58 – 65. [5] Amsel S, Maniatis A, Tavassoli M, Crosby WH. The significance of intramedullary cancellous bone formation in the repair of bone marrow tissue. Anat Rec 1969;164:101 – 12. [6] Patt HM, Maloney MA. Bone marrow regeneration after local injury: a review. Exp Hematol 1975;3:135 – 48. [7] Bab IA. Postablation bone marrow regeneration: an in vivo model to

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24] [25] [26] [27]

study differential regulation of bone formation and resorption. Bone 1995;17:437S – 41S. Liang CT, Barnes J, Seedor JG, Quartuccio HA, Bolander M, Jeffrey JJ, et al. Impaired bone activity in aged rats: alterations at the cellular and molecular levels. Bone 1992;13:435 – 41. Suva LJ, Seedor JG, Endo N, Quartuccio HA, Thompson DD, Bab I, et al. Pattern of gene expression following rat tibial marrow ablation. J Bone Miner Res 1993;8:379 – 88. Tanaka H, Barnes J, Liang CT. Effect of age on the expression of insulin-like growth factor-I, interleukin-6, and transforming growth factor-h mRNAs in rat femurs following marrow ablation. Bone 1996;18:473 – 8. Ishizaka M, Tanizawa T, Sofue M, Dohmae Y, Endo N, Takahashi HE. Bone particles disturb new bone formation on the interface of the titanium implant after reaming of the marrow cavity. Bone 1996; 19:589 – 94. Hara T, Hayashi K, Nakashima Y, Kanemaru T, Iwamoto Y. The effect of hydroxyapatite coating on the bonding of bone to titanium implants in the femora of ovariectomised rats. J Bone Joint Surg 1999;81-B:705 – 9. Schmidmaier G, Wildemann B, Schwabe P, Stange R, Hoffmann J, Sudkamp NP, et al. A new electrochemically graded hydroxyapatite coating for osteosynthetic implants promotes implant osteointegration in a rat model. J Biomed Mater Res 2002;63:168 – 72. Kuroda S, Virdi AS, Li P, Healy KE, Sumner DR. A low temperature biomimetic calcium phosphate surface enhances early implant fixation in a rat model. J Biomed Mater Res 2004;70A:66 – 73. De Ranieri A, Virdi AS, Kuroda S, Healy KE, Sumner DR, Saline irrigation does not affect bone formation or implant fixation strength in a rat model. J Biomed Mater Res [In press]. Clarke SA, Brooks RA, Lee PT, Rushton N. Bone growth into a ceramic-filled defect around an implant. The response to transforming growth factor beta1. J Bone Joint Surg, Br 2004;86:126 – 34. Turner TM, Sumner DR, Urban RM, Rivero DP, Galante JO. A comparative study of porous coatings in a weight-bearing total hiparthroplasty model. J Bone Joint Surg 1986;68-A:1396 – 409. Sumner DR, Turner TM, Urban RM, Leven RM, Hawkins M, Nichols EH, et al. Locally delivered rhTGF-h2 enhances bone ingrowth and bone regeneration at local and remote sites of skeletal injury. J Orthop Res 2001;19:85 – 94. Bernhardt R, Scharnweber D, Muller B, Thurner P, Schliephake H, Wyss P, et al. Comparison of microfocus- and synchrotron X-ray tomography for the analysis of osteointegration around Ti6Al4Vimplants. Eur Cell Mater 2004;7:42 – 51. Sumner DR, Turner TM, Cohen M, Losavio P, Urban RM, Nichols EH, et al. Aging does not lessen the effectiveness of TGF-h2 enhanced bone regeneration. J Bone Miner Res 2003;18:730 – 6. Lind M, Overgaard S, Nguyen T, Ongpipattanakul B, Bu¨nger C, Søballe K. Transforming growth factor-h stimulates bone ongrowth: hydroxyapatite-coated implants studied in dogs. Acta Orthop Scand 1996;67:611 – 6. Eckhoff DG, Turner AS, Clarke R, Hastings R, Nguyen T, Cromwell M, et al. Dose effect of rhTGF-h1 on bone growth around prosthetic implants in an age-controlled model. Trans ORS 1996;21:95 [Abstr]. Schmidmaier G, Wildemann B, Gabelein T, Heeger J, Kandziora F, Haas NP, et al. Synergistic effect of IGF-I and TGF-beta1 on fracture healing in rats: single versus combined application of IGF-I and TGFbeta1. Acta Orthop Scand 2003;74:604 – 10. Geesink RG. Osteoconductive coatings for total joint arthroplasty. Clin Orthop 2002;395:53 – 65. LeGeros RZ. Properties of osteoconductive biomaterials: calcium phosphates. Clin Orthop 2002;81 – 98. Dupont WD, Plumer Jr WD. Power and sample size calculations. A review and computer program. Control Clin Trials 1990;11:116 – 28. De Ranieri A, Virdi AS, Kuroda S, Shott S, Dai Y, Sumner DR, Local application of rhTGF-h2 modulates dynamic gene expression in a rat implant model. Bone [Epub ahead of print March 24 2005].