Intermittent PTH administration and mechanical loading are anabolic for periprosthetic cancellous bone

Abstract

The purpose of this study was to determine the individual and combined effects on periprosthetic cancellous bone of intermittent PTH (iPTH) and mechanical loading at the cellular, molecular, and tissue levels. Porous titanium implants were inserted bilaterally on the cancellous bone of adult rabbits beneath a loading device attached to the distal lateral femur. The left femur received a sham loading device. The right femur was loaded daily, and half of the rabbits received daily PTH. Periprosthetic bone was evaluated up to 28 days for gene expression, histology, and µCT analysis. Loading and iPTH increased bone mass by a combination of two mechanisms: 1) altering cell populations in a pro-osteoblastic/anti-adipocytic direction, and 2) controlling bone turnover by modulating the RANKL-OPG ratio. At the tissue level, BV/TV increased with both loading (+53%, p<0.05) and iPTH (+54%, p<0.05). Combined treatment showed only small additional effects at the cellular and molecular levels that corresponded to a small additive effect on bone volume (+13% compared to iPTH alone, p>0.05). This study suggests that iPTH and loading are potential therapies for enhancing periprosthetic bone formation. The elucidation of the cellular and molecular response may help further enhance the combined therapy and related targeted treatment strategies.

1 Introduction

The number of total arthroplasties performed annually in the United States exceeds 1 million and is expected to rise through the next decade^1,2. The amount of bone surrounding these implants is important for the long-term success of total joint replacement. Periprosthetic fractures and implant loosening due to unsuccessful osseointegration are two modes of failure for total joint arthroplasty that depend on the amount of the surrounding cancellous bone^3–6. Enhancement of this periprosthetic bone is critical to long-term fixation and ultimate clinical success of a joint replacement surgery^3–6.

Both mechanical loading and intermittent parathyroid hormone administration (iPTH) are proven anabolic therapies in bone and are potential candidates for enhancing periprosthetic bone formation. In both animal and human studies, investigators have established a clear anabolic effect of mechanical loading in both cortical and cancellous tissue^7–11. Numerous histologic studies show that loading increases osteoblast numbers and decreases osteoclast activity leading to increased bone mass^9,12–14. Similarly, in pre-clinical and clinical studies, iPTH administration increases bone mass in both cortical and cancellous bone. In scenarios of low bone mass, the anabolic effects of iPTH occur primarily through increased numbers of matrix-synthesizing osteoblasts¹⁵. The origin of this increase in osteoblast number is likely multi-factorial, with increased osteoblastogenesis, decreased osteoblast apoptosis, and activation of quiescent lining cells as probable contributing factors^15–17. Osteoclast number, and hence activity, is increased by iPTH, most likely through the production of RANKL by stromal cells in the marrow environment. This enhancement in osteoclast activity may be necessary for the overall anabolic effects of iPTH^18,19. Intermittent administration of teriparatide, a recombinant form of PTH, is an FDA-approved treatment for osteoporosis, and currently the only approved skeletal anabolic pharmacologic agent on the market in the US^20–22. The few studies that looked at the effects of iPTH on osseointegration into implants found differences between periprosthetic and generalized bone responses^23,24. Thus, the effects of iPTH on periprosthetic bone may be unique and the underlying mechanisms, which are largely unknown, cannot be extrapolated from non-implant, and non-surgical, skeletal studies^25,26.

In addition to an incomplete picture of the effects of iPTH on periprosthetic cancellous bone, our understanding of the effects of iPTH combined with mechanical loading is limited, and, to our knowledge, has not been examined previously in periprosthetic bone. In a previous study, we examined the effects of iPTH and loading in a similar model, but one lacking an implant²⁵. The current study expands upon that work by utilizing a more clinically relevant porous titanium implant, and by performing a more in depth examination of the molecular and cellular response. The purpose of this study was to determine the effects of combined mechanical loading and iPTH on periprosthetic cancellous bone beneath a porous titanium implant, and to examine these effects at the cellular and molecular levels.

2. Methods

2.0 Antibodies and culture reagents

SP1.D8, a rat monoclonal antibody to pro-collagen type 1 (PC-1) was obtained from Developmental Studies Hybridoma Bank (Iowa City, IA) and used at 1:50. Anti-proliferating cell nuclear antigen (PCNA), sc-56, was obtained from Santa Cruz (Santa Cruz, CA) and used at 1:100. The TRAP staining kit (387-A), which allows quantitation of osteoclast number, was purchased from Sigma-Aldrich (St Louis, MO).

2.1 Experimental Design

This study used a well established rabbit model to investigate the effects of mechanical loading and iPTH on periprosthetic bone and examine the underlying cellular mechanisms^8,25,27. A porous implant and loading device were implanted bilaterally in the distal femurs of skeletally mature (31 weeks), male New Zealand white rabbits (n=125) as described previously^27,28 (Figure1A). Cylindrical, 5mm diameter, 2mm in length, porous implants (Orchid Orthopedic Solutions, Holt, MI) were used in the study. The porous implant was a monoblock of porous titanium foam with a µm cell diameter of ~900, an interconnecting pore diameter of ~300µm (MVIL of 565.1 ± 170.6 µm), and an average porosity of 74.4%. Mean void intercept length and porosity measurements were determined by 2D metallography techniques. The implants were packaged and gamma sterilized (25–45 kGy). The pore dimensions are comparable to porous –coated clinical implants.

Figure 1.

A. A porous titanium implant. B. Picture of loading device and implant in place on the distal lateral femur of a rabbit. X’s represent periprosthetic bone of interest. C. Histological areas of interest of bone core (white boxes), shown on a microCT cross-section of the core.

(Figure 1A). The right femur received a functional loading device, and the left femur a sham device as loading control.

Gene expression, cellular responses, and bone morphology were analyzed in the periprosthetic tissue using qPCR, immunohistochemical and histological staining, and microcomputed tomography (microCT), respectively. Animals used for qPCR (n=7–10/group/time point, 100 samples total, 50 rabbits) were euthanized at the 3, 7, and 14 day time points. Animals for immunohistochemistry and histological staining (n=7/group/time point, 112 samples total, 56 rabbits total) were euthanized at 3, 7, 14 and 28 days. Rabbits (n=9–10/group, 38 samples total, 19 rabbits) were euthanized at 28 days for microCT analysis. All rabbits were euthanized using intravenous Sleepaway (Sodium Pentobarbital 26%/Isopropyl alcohol 10%, 2.0 ml) euthanasia solution (Zoetis, Florham Park, NJ).

2.2 Surgical Procedure

The surgical procedure described previously was modified for this study²⁷. Surgical anesthesia was achieved with 0.05 mg/kg atrophine sulfate, 35 mg/kg ketamine hydrochloride, and 0.5 mg/kg actylpromazine injected intramuscularly and maintained by isoflurane inhalation. The femoral condyle was exposed with a standard posterior lateral surgical approach. Once exposure was achieved, the layer of cortical bone was removed to a depth of 1 mm with a milling guide and custom routing device. Following milling, the guide was removed, and the implant was placed on the exposed trabecular bone. With the implant in place, the stationary baseplate of the in vivo loading device was secured over the implant with two bicortical screws (Figure 1B). Once the loading device was in place, the device on the right limb was locked to prevent inadvertent loading of the porous implant. The left sham loading device was permanently locked by design. The surgical procedure required 60 minutes. Ampicillin (25 mg/kg) was used as antibiotic prophylaxis, and buprenorphine (0.05 mg/kg) was used as postoperative analgesia.

2.3 Loading Regimen and Parathyroid Hormone Dosing

Loading began one day after surgery for all rabbits. Rabbits were placed in a customized restraining cage that allowed access to the right femur. After manually unlocking the loading device, a compressive load (1 MPa, 1 Hz, 50 cycles/day, 5 days/week) was applied to the top surface of the implanted device. Specifically, a load of 18N was applied to the loading device. Based on the 5 mm loading core diameter, this load corresponds to an apparent stress of approximately 1 MPa⁸. The compressive load was applied using an automated system that utilized an inline load cell (ELFS-T3E-20L, Measurement Specialties, Fairfield, NJ) with load-based feedback control²⁷. In previous studies, this loading regimen was shown to maximize cancellous bone formation in native cancellous tissue while corresponding to physiologic loads in this model⁸. Between loading sessions, rabbits were allowed unrestricted cage activity. Locking of the device in between loading sessions ensured that unintentional loading did not occur. All rabbits were loaded on the day of euthanasia, 4 hours prior to euthanasia administration.

Parathyroid hormone (hPTH1-34, 20 µg/kg, SQ, 5 days/week, Lilly, Indianapolis, Indiana) was administered starting one day after surgery and continued for the duration of the experiment in half of the rabbits while the others received saline injections. For dynamic histomorphometry, rabbits were injected with calcein (15 mg/kg) at 14 and 4 days before euthanasia. All in vivo procedures, loading regimens, injections, and euthanasia were approved by the IACUC of the Hospital for Special Surgery.

2.4 Cancellous Core Preparation

For microCT, immunohistochemistry and enzymatic staining, distal femurs were removed bilaterally after euthanasia, fixed in hardening resin and an 8 mm core was removed with a coring device (Core Drill 101055, Starlite Industries, PA) centered around the porous implant. For immunohistochemistry and enzymatic staining, 8 mm diameter cancellous cores were fixed in 10% neutral buffered formalin for 72 hours and then decalcified using 10% ethylenediaminetetraacetic acid for 4 weeks. Following decalcification, the porous implants were removed from the core, the cores were cut in half parallel to the long axis of the cylinder, and both halves were embedded in paraffin. The cores for microCT were placed in 70% ethanol. Tissue cores for quantitative PCR (qPCR) were obtained manually with a tissue biopsy punch (4 mm diameter, 3 mm depth) after removing the overlying metal implants, cleaned of surrounding soft tissue, flushed with sterile saline to remove marrow, snap-frozen in liquid nitrogen and stored at −80°C until processed for RNA analysis.

2.5 qPCR

Frozen bone samples were pulverized using an oscillating mill (MM400, Resch, Haan, Germany), and RNA was extracted using TRIzol (Invitrogen #14 596-026, Grand Island, New York) and RNeasy Mini Kits (Qiagen #74105, Hilden, Germany), including a DNase removal step. The concentration of RNA was determined (Nano Drop, ND-2000, Wilmington, DE), and RNA integrity (RIN value) was determined for each sample (BioAnalyzer, Agilent Technologies, Inc., Santa Clara, CA). RNA was converted to cDNA in batches of 1 ug (High Capacity cDNA Reverse Transcription kit and GeneAmp PCR System 2700, Applied Biosystems, Foster City, CA). Gene expression was analyzed by qPCR (Opticon Monitor 3, Bio-Rad, Hercules, CA) using previously tested primer probes when available or custom designed probes (Table 1), and a custom master mix utilizing platinum Taq DNA polymerase (10966-026, Invitrogen, Grand Island, NY). Primer sensitivity and specificity was confirmed for all genes using submarine agarose elctrophoresis to size PCR products. For newly designed primers, the band was excised, and the DNA eluted (QIAquick gel extraction kit, Qiagen #28704, Hilden, Germany) and sequenced. Genes were selected based on their importance as markers of relevant cellular changes and/or signaling activity. Collagen 1a1, Runx2, and osteocalcin are markers of osteoblast differentiation. BMP-2 is implicated as an important signaling ligand in osteoblast differentiation and bone formation. PPARγ is the key transcription factor for adipogenesis. The cytokine RANKL is required for osteoclast differentiation and activation; OPG acts as its decoy. β-catenin is a key intracellular mediator of Wnt signaling^29,30, and Sclerostin, encoded by the SOST gene, is a Wnt signaling antagonist.

Table 1.

Primers used for real-time qPCR

Gene	Source	Forward Primer (5’ to 3’)	Reverse Primer (5’ to 3’)	Annealing Temperature
18s	Monjo et al., 2008	GTAACCCGTTGAACCCCATT	CCATCCAATCGGTAGTAGCG	57.0
Runx2	Monjo et al. 2008	GAAGCCCAGCGGTGCA	CACTACCTCGCTGCCCTCC	52.8
Collagen 1a1	Monjo et al. 2008	AGAGCATGACCGATGGATTC	CCTTCTTGAGGTTGCCAGTC	52.5
Osteocalcin	Monjo et al. 2008	GAAGCCCAGCGGTGCA	CACTACCTCGCTGCCCTCC	58.5
PPARγ	Custom	CACTCCTTTGACATCAAGCC	TGATTGCACTTTGATACTCCTG	59.0
BMP-2	Tang et al. 2011	TTGGAGGAGAAGCAAGGTGT	TACGAGCAAAGGCCTGATTT	52.5
SOST	Custom	ATGACGCCACGGAAGTTATC	CGGTTCATGGTCTTGTTCTC	52
RANKL	Custom	ACACCCTCATGAAGGGAGGAAGC	GGAAGGGTTGGACACCTCCACA	51
OPG	Custom	GGAAACAGTGAATCAACTCAAA	TCTCTACACTCTCAGCATTTACTT	57.2
β-catenin	Custom	AAATCTTGCCCTTTGTCC	CAACCTTCCACTATTTCTTCC	60

2.6 Immunohistochemistry and Histologic Staining

Immunohistochemical staining of the periprosthetic bone was performed for procollagen 1, a marker of active osteoblasts, and PCNA, a marker of cell division. TRAP staining was used to identify the number of osteoclasts. Sections were deparaffinized, rehydrated, and placed in hot citrate buffer for 30 mins (PC-1) or 1 hour (PCNA) for antigen retrieval. The remainder of the protocol for immunohistochemical staining was outlined in a previous study²⁵. TRAP staining was performed according to the manufacturer’s instructions, and sections were counter-stained with methyl-green, dehydrated, and cover-slipped. Sections that underwent TRAP staining were also used to quantify adipocyte density because this protocol allowed the most favorable bone marrow preservation.

Quantification of staining was performed using histology software (Bioquant, Nashville, Tennessee). Specificity of immunohistochemistry antibody staining was confirmed using positive and negative controls. For PC-1, TRAP, and adipocyte density, the entire loading surface (5 mm) to a depth of 1 mm was analyzed. PC-1 stain was used to quantify percent of viable osteoblasts per total bone surface (Ob.S/BS). Adipocyte density (N. Ad/Ma. A) was determined as the number of apparent adipocytes (measured as adipocyte ghosts) per total bone marrow area (area between trabecular bone in mm²). The number of osteoclasts per bone surface (N. Oc/BS) was determined using TRAP staining of multi-nucleated cells along the bone surface. Three 1 mm × 0.6 mm boxes along the implant surface were used for PCNA analysis (Figure 1C). The percent osteoblasts positive for PCNA (Positive Ob/Total Ob) was used to quantify the number of osteoblasts expressing proliferative activity.

2.7 Microcomputed Tomography

Eight mm diameter cores of the peri-implant region were used for trabecular bone microCT analysis (µCT 35 system, Scanco Medical, Bassersdorf, Switzerland). Scans were performed in 70% ethanol using a 15 µm³ voxel size, 55 kVp, 145 µA, 0.36 degree rotation step (360° angular range), and 400 ms exposure per view. Scanco microCT software (HP, DECwindows Motif 1.6) was used for 3D reconstruction and image viewing. After 3D reconstruction, volumes were segmented using a global threshold of 520 mg HA/cm³. Bone volume fraction (BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th), and trabecular separation (Tb.Sp) were calculated for a cylindrical volume of interest (d = 4.08 mm; h = 0.5 mm) centered directly under the implant surface.

2.8 Statistical Analysis

Summary statistics were calculated for all outcome measures including mean, standard deviation and 95% confidence intervals (95% CI). Generalized estimating equation (GEE) models were used to assess the main effects and interaction of PTH treatment and loading condition on each outcome measure. Sample size reduced the power and prevented incorporation of time as an additional variable, although qualitative changes ware noted when appropriate. Parameter estimates of the models were reported using beta coefficient estimates with standard deviations (SD) and 95% confidence intervals also calculated. GEE models were fitted using PROC GENMOD with REPEATED statements in SAS 9.2 (SAS Institute, Cary, NC) and statistical significance set to alpha equal to 0.05.

3. Results

3.1 Additive effect of mechanical loading and iPTH on osteoblast and adipocyte responses

A decrease in Ob.S/BS for all groups from Day 3 compared to Day 28 was observed (Figure 2A). Intermittent PTH delayed this decrease, resulting in significantly higher osteoblast activity at Day 14 (Figure 2A, 3A). The effect of mechanical loading was similar, although more modest (Figure 2A, 3A). To determine the mechanism behind the delay in decreased osteoblast surface, we measured the number of cells exhibiting PCNA expression^17,31,32. This index increased significantly with both loading and iPTH treatment (Figure 2B, Figure 3B). Individually, the effects of iPTH and mechanical loading on adipocyte density (N. Ad/Ma.V) were similar: both treatments significantly decreased adipocytes by Day 28 (Figure 2C). As found with Ob.S/BS, the combined effects of the treatments on adipocyte density (N. Ad/Ma.V), were additive, but not synergistic.

Figure 2.

Mechanical loading and iPTH stimulate proliferation and activity of osteoblasts. Connected lines represent significant changes with iPTH, and non-connected lines indicate significant changes with mechanical loading (both present, as in A, B, and C, represent significant changes with both treatments). A. Changes in osteoblast surface/total bone surface (Ob.S/BS) with treatments and time as indexed by pro-collagen I staining. B. Changes with treatments and time in percentage of osteoblasts expressing PCNA. C. Changes in adipocyte number with treatments and time per total bone marrow area in mm² (N. Ad/Ma.Ar).

Figure 3.

Representative images from immunohistochemistry and histology showing changes in osteoblast number and osteoblast DNA repair along the periprosthetic surface. White dotted lines represent location of bone-implant interface. A. Mechanical loading and iPTH increased osteoblasts (Day 14). B. Mechanical loading and iPTH increased PCNA positive osteoblasts (Day 14). C. Fluorescence imaging with calcein labeling showing rapid bone formation in the periprosthetic region. The image to the right is a similarly scaled microCT image for reference (not the exact same cross section).

Changes at the molecular level reflected those seen for osteoblasts and adipocytes, with a few important exceptions. Both mechanical loading and iPTH reduced PPARγ mRNA expression (Figure 4A), confirming our earlier findings of reduced adipocyte number. Similarly, they significantly increased BMP-2 expression, although at different time points (Figure 4B, 4C). For mechanical loading, an increase in Runx2 supported an increase in osteoblast differentiation (Figure 4D). Decreased mRNA levels for Runx2, Collagen 1a1, and Osteocalcin, following iPTH treatment were unanticipated. These three markers of osteoblast differentiation decreased significantly with iPTH therapy at multiple time points, with the most significant changes at Day 7 (Figure 4E–G, Table 2).

Figure 4.

Mechanical loading and iPTH alter markers of osteoblast or adipocyte differentiation. Connected lines indicate significant changes with iPTH (A, C, E, F, G), and non-connected lines indicate significant changes with mechanical loading (A, B, D). Only time points with statistically significant changes are shown, excluding: Col-1a1 Day 3; Osteocalcin Day 3, Day 14; and PPARγ Day 3 (See Supplementary Data 1).

Table 2.

Changes in gene expression measured by qPCR

	Day 3					Day 7					Day 14
	VN	VL	PN	PL	Sig.	VN	VL	PN	PL	Sig.	VN	VL	PN	PL	Sig.
Runx2	0.62 (0.43)	0.60 (0.49)	0.38 (0.24)	0.43 (0.31)		2.4 (1.1)	2.2 (0.92)	0.69 (0.30)	1.2 (0.45)	PTH	0.67 (0.26)	1.1 (0.43)	0.68 (0.38)	0.86 (0.52)	Load
Col-1a1	2.7 (1.1)	3.1 (2.6)	1.6 (0.90)	2.0 (0.54)	PTH	5.8 (2.1)	7.0 (3.9)	3.7 (1.2)	3.5 (2.0)	PTH	3.8 (2.8)	4.5 (2.7)	4.3 (1.5)	3.5 (2.0)
Osteocalcin	3.6 (1.9)	3.3 (2.9)	2.02 (1.1)	2.3 (1.2)	PTH	11.6 (5.4)	11.6 (4.2)	5.5 (2.4)	6.9 (2.7)	PTH	8.7 (4.1)	9.45 (2.6)	5.7 (2.1)	7.8 (2.9)	PTH
PPARγ	1.9 (0.9)	3.3 (1.9)	3.8 (3.1)	3.4 (1.8)	Load	0.75 (0.35)	0.73 (0.34)	0.97 (0.94)	0.55 (0.20)		4.3 (2.2)	2.3 (1.2)	1.8 (1.4)	1.7 (0.78)	Load PTH
BMP-2	3.7 (1.9)	7.1 (3.3)	5.8 (2.8)	8.2 (6.0)	Load	3.2 (0.57)	3.72 (2.2)	3.4 (1.6)	3.0 (1.3)		3.6 (2.2)	4.4 (0.90)	6.7 (1.9)	4.7 (2.1)	PTH
SOST	2.1 (1.8)	3.4 (1.8)	2.9 (1.7)	3.6 (2.8)		3.4 (2.2)	2.35 (0.98)	1.4 (0.5)	1.7 (0.79)	PTH	6.5 (3.5)	6.1 (3.5)	5.8 (3.3)	5.8 (3.2)
RANKL	6.2 (5.1)	7.8 (6.1)	7.8 (5.7)	8.7 (6.8)		2.5 (0.67)	2.1 (0.68)	4.0 (1.7)	3.1 (1.3)	PTH	2.6 (1.1)	4.3 (2.1)	4.6 (1.7)	3.0 (1.0)	Load PTH
OPG	6.7 (2.6)	11.9 (4.8)	8.9 (5.2)	15.2 (8.2)	Load	3.9 (1.5)	4.0 (1.2)	4.5 (2.4)	3.5 (2.0)		3.8 (1.6)	5.7 (3.9)	6.3 (3.9)	5.8 (2.1)
RANKL:OPG	0.96 (0.69)	0.64 (0.30)	0.97 (0.61)	0.51 (0.43)		0.72 (0.33)	0.55 (0.27)	1.0 (0.28)	1.0 (0.34)		0.73 (0.24)	0.85 (0.46)	0.90 (0.38)	0.54 (0.23)
Β-catenin	5.2 (1.7)	8.6 (4.1)	8.11 (4.1)	6.7 (3.4)	Load	4.7 (1.0)	4.6 (1.8)	5.7 (2.4)	3.5 (1.4)		3.4 (1.5)	3.8 (2.5)	6.0 (2.4)	4.9 (2.3)	PTH

VN: Vehicle No Load, VL: Vehicle Load, PN: PTH No Load, PL: PTH Load. All units are a ratio of Relative mRNA/18s mRNA

3.2 Mechanical loading and PTH alter osteoclast number in opposite directions and at different time points

The osteoclast response to mechanical loading and iPTH differed in both direction and timing. Loading suppressed osteoclast number only at Day 7 (Figure 5A). In contrast, the effects of iPTH on osteoclast number were seen solely at 28 days (Figure 5A).

Figure 5.

Mechanical loading and iPTH alter osteoclast number. Connected lines indicate significant changes with iPTH (A[Day 28], C), and non-connected lines indicate significant changes with mechanical loading (A[Day 7], B). A. Changes in osteoclast number per bone surface in mm (N.Oc./BS) with treatment and time. B. Mechanical loading increased OPG mRNA expression at Day 3. C. iPTH increased RANKL mRNA expression at Day 7. D and E. Mechanical loading and iPTH led to non-significant changes in the RANK:OPG ratio.

The osteoclast responses were supported by changes in OPG and RANKL mRNA expression. Paralleling the early decrease in osteoclast number observed with mechanical loading, we saw a significant increase in OPG in the loaded groups at the 3 day time point (Figure 5B) while iPTH therapy had no impact at any time point. In contrast, iPTH increased RANKL expression at a later time point (Day 7) (Figure 5C). Changes in RANKL:OPG ratio showed trends toward a decreased ratio at Day 3 with loading, and increased at Day 7 with iPTH, but did not reach significance (Figure 5D–E).

3.3 Mechanical loading and iPTH regulate Wnt/β-catenin signaling

Osteoblast formation and function are regulated by a number of factors that converge on Wnt/ β-catenin, signaling³³. A key effector for these signals is SOST, the gene encoding sclerostin, which was downregulated with iPTH administration at Day 7, consistent with increased osteoblast activity at Day 14 (Figure 6A). The changes in SOST expression were not significant for the loaded groups (Figure 6A). Levels of β-catenin mRNA were also measured. Mechanical loading induced a significant increase in β-catenin levels at Day 3, although an interaction with iPTH antagonized this loading effect (Figure 6B). At Day 14, iPTH treatment led to significant increases in β-catenin (Figure 6C). Although both loading and iPTH are thought to act, at least in part, through Wnt signaling, we saw no synergistic effect at any time point for either gene. Moreover, in all cases in which we saw an effect from iPTH, the combined treatment group had no greater effect than the iPTH group alone (Figure 6A, C).

Figure 6.

Mechanical loading and iPTH alter Wnt signaling. Connected lines indicate significant changes with iPTH (A, C), and non-connected lines indicate significant changes with mechanical loading (B). A. iPTH decreased SOST mRNA at Day 7. B. Mechanical loading increased β-catenin mRNA at Day 3, but this effect was antagonized when combined wtih iPTH treatment. C. iPTH increased β-catenin mRNA at Day 7.

3.4 Combined treatment increases periprosthetic bone volume only minimally compared to individual treatment

MicroCT analysis of periprosthetic bone revealed significantly increased bone volume fraction as a result of both mechanical loading and iPTH (Table 3). The large increase in BV/TV (53%) in response to mechanical loading was attributable mainly to a 50% increase in trabecular thickness (Table 3). Administration of iPTH induced a very similar outcome in BV/TV (+54%, Table 3). Trabecular separation decreased 33%, and there was a trend toward increased trabecular number with iPTH therapy (+25%, p =0.096). The combined treatment group showed a small additive effect, with a further increase of 13% compared to the individually treated groups, and 74% compared to the vehicle control group (Table 3).

Table 3.

Changes in bone morphology measured through µCT (Day 28)

Group	BV/TV	Tb.N	Tb.Th (mm)	Tb.Sp (mm−1)
Vehicle No Load	0.27 (0.14)	5.89 (2.09)	0.10 (0.02)	0.22 (0.07)
Vehicle Load	0.41 (0.11)	5.98 (1.21)	0.15 (0.02)	0.19 (0.06
iPTH No Load	0.42 (0.11)	7.32 (1.86	0.11 (0.03)	0.15 (0.06)
iPTH Load	0.47 (0.14)	7.66 (1.86)	0.12 (0.02)	0.14 (0.06)
Significance	PTH (p=0.007), Load (p=0.005)	None (PTH p= 0.096)	Load (p<0.0001)	PTH (p=0.034)

BV/TV = Bone Volume/ Total Volume, Tb.N = Trabecular Number, Tb.Th = Trabecular Thickness, Tb.Sp = Trabecular Separation

4. Discussion

A robust anabolic skeletal response in the periprosthetic tissue greatly diminishes the chances of implant loosening and periprosthetic fracture, and ultimately leads to an increase in the lifespan of total joint replacement prostheses. Both mechanical loading and intermittent parathyroid hormone therapy are anabolic to bone, but their role in periprosthetic bone and their effects as a combination therapy are unknown. In the present study, we found strong individual treatment effects on bone volume, and a small additive relationship with the combined treatments at the immediate periprosthetic surface. These effects were mediated by pro-osteoblastic/anti-adipocytic molecular and cellular responses, with separate control of osteoclast activity by the two treatments through changing RANKL-OPG levels.

Previously, our laboratory developed a rabbit in-vivo model of mechanical loading in which cancellous bone exhibits an anabolic functional adaptation^8,28. We then used the same model to examine the effects of mechanical loading combined with iPTH on cancellous bone in the absence of a titanium implant²⁵. The current study is unique for a number of reasons: 1) the presence of a porous implant allowed us to examine the bone response to loading and iPTH using a more clinically relevant prosthesis, 2) the region of cancellous bone examined was closer to the implant surface, and 3) we documented the accompanying molecular and cellular responses for the first time.

The effects of iPTH and mechanical loading on adipocyte and osteoblast responses in periprosthetic cancellous bone are consistent with current evidence from non-implant studies. The mesenchymal stem cell is the common precursor cell to osteoblasts and adipocytes, and stimulation of these precursor cells toward the osteoblast lineage plays an important role in the anabolic bone response³⁴. Our data support this view in that molecular markers of adipogenesis and some markers of osteoblastogenesis are decreased or increased, respectively. These findings are accompanied by increased numbers of active osteoblasts and decreased adipocyte density. One notable exception was related to PTH-based inhibition of the osteoblast markers Runx2, Collagen 1a and Osteocalcin. While contradictory to common dogma regarding osteoblast markers, others have reported similar changes²⁹. Specifically, Tu et al.²⁹ suggested that decreased mRNA levels may reflect a slowed osteoblast maturation process in the vehicle group when compared to the iPTH group leading to an accumulation of osteoblast markers. Given the findings by Tu et al., and our observation of this iPTH effect, expression of pre-osteoblast markers in-vivo may be more complex than previously thought.

While our results suggest that mesenchymal stem cell differentiation is likely an important contributing factor to the observed increased osteoblastic activity, other mechanisms have been studied. Activation of quiescent bone lining cells plays an important role in the osteoblast response to iPTH^16,21. Further, increasing the lifespan of active osteoblasts is another hypothesized mechanism³⁵. For example, PCNA, although commonly used as a marker of cell proliferation, also correlates with an anti-apoptotic state relating to its role in the DNA repair process^17,31,32. We examined expression of PCNA in mature osteoblasts, which have minimal proliferation activity, and observed increased PCNA positive osteoblasts with both treatments individually and a strong additive effect in the combined group (+24% over PTH alone). Increased osteoblast lifespan due to enhanced DNA repair may play an important role in increasing osteoblast number under combined treatments with iPTH and mechanical loading.

In vitro and in vivo evidence suggests that iPTH increases osteoclast number and hence overall activity, an outcome that likely results in increased bone turnover accompanying this treatment¹⁸. The effects of mechanical loading are less clear, but many studies have observed decreased osteoclast numbers with treatment^12,13. In periprosthetic bone, we found that these same trends of increased osteoclast number with iPTH and decreased number with loading are maintained, but differ in timing and the molecular factors responsible for the changes. Thus, mechanical loading led to an early decrease in osteoclast number through an increase in OPG. In contrast, iPTH increased RANKL mRNA levels, and increased osteoclast number. The importance of the increase in osteoclast number with iPTH treatment in the net anabolic response is controversial^36–38. Osteoclasts may be critical to the increased osteoblast response, and an essential coupling between the two processes has been described^18,19. However, the increased osteoclast response may simply be a consequence of increased bone turnover^38,39. In the latter scenario, blocking osteoclast activity, and leaving the osteoblast response unchecked, may enhance bone formation.

Our study had the advantage of comparing therapies that increased (iPTH) and decreased (mechanical loading) osteoclast numbers. However, many questions remain regarding the net effect of this dual treatment strategy. For example, the additional effect on BV/TV of the combined treatment group could be a consequence of reduced osteoclast numbers due to mechanical loading, or simply a result of further enhanced osteoblast activity. Alternatively, by reducing osteoclasts, mechanical loading may disrupt the coupling mechanism, and lead to a dampened response of iPTH. This latter hypothesis may explain the relatively weak additional effect on bone volume of the combined group compared to the individual alone. Results from studies that make use of anti-osteoclast agents such as bisphosphonates or denosumab (a monoclonal antibody targeting RANKL) in combination with iPTH have supported both scenarios^39–44. The answer may depend on the available re-modeling space. For example, some studies suggest that in situations of healing defects in which this space is abundant, osteoclast activity is not required^36,45. This may be the case for our model.

The periprosthetic changes at the tissue level were clear and reflected the cellular and molecular changes. We chose to examine the most immediate periprosthetic surface because this bone is most critical to implant survival^3,4,46, and we expected this area to have the largest cellular response. Calcein labeling showed this rapid bone formation at the immediate implant interface (Figure 3C). Indeed, bone volume changes at 0.5 mm showed a strong increase from individual treatments (iPTH: +54%, p<0.05; Loading: +53%, p<0.05). Compared to the previous study by Fahlgren et al. which examined a generalized bone response to PTH (+30%) and loading (+10%) without the presence of an implant, we saw a greater response in the periprosthetic bones examined in this study. This difference may be attributed to the presence of the porous implant.²⁵ In addition, combined treatment showed a small additional effect (+13% compared to iPTH alone, p>0.05; +74% compared to vehicle control, p<0.05). This small additional combined effect may not be surprising considering the molecular changes of individual genes showed little evidence for synergy. For example, PPARγ, the central regulator of adipogenesis, which changed with individual treatments, showed no additional effect with combined therapy. In addition, the Wnt signaling pathway, thought to play a major role in mediating the anabolic response of both therapies, gave no evidence for a strong combined effect. Careful examination of the literature reveals that while many studies see a synergistic effect of the combined treatments on bone formation rate^47–49, none have observed a synergistic effect on cancellous bone volume^25,49,50. Therefore, a synergistic effect of iPTH and mechanical loading is unlikely for cancellous bone, and our data demonstrate that this result holds for periprosthetic cancellous bone as well.

Although not synergistic, the additional effect from the combined groups may translate to a clinically beneficial improvement worth pursuing. The magnitude of periprosthetic bone enhancement that correlates with clinical improvements is not well known. In addition, the results of this investigation revealed that achieving optimal combined effects from these treatments might be more complicated than originally considered. For example, we observed that anti-osteoclastic loading effects and pro-osteoclastic iPTH effects were activated at different time points during our study (Days 7 and 3, respectively). In addition, the individual treatments exerted temporally different effects on β-catenin and BMP-2 expression, with a loading response apparent at day 3 and that of iPTH detectable at day 14. While β-catenin is typically thought to regulate bone formation post-translationally, several studies suggested that increased β-catenin gene expression can be indicative of increased Wnt signaling^29,30. Thus, for enhancement of periprosthetic bone, the timing of administration of multiple anabolic therapies may be critical, and further studies may be necessary to investigate these findings.

In this study, we focused on the response of periprosthetic bone to these anabolic therapies because formation of new bone is likely to play a critical role in determining successful implant fixation^3,4. We observed a strong individual effect of iPTH and mechanical loading, and a small additional effect from combined treatment, mediated through a pro-osteoblastic/anti-adipocytic response, and control of the osteoclast response through RANKL-OPG changes. The results of the paper also suggest that small regions adjacent to the implant may play the most important role in implant stability. This study did not examine the extent of bone ingrowth into the porous implant. While relevant to implant stability, we chose to focus on the response of periprosthetic tissue, and examine these issues separately. Additional animals and further experimentation with higher resolution techniques will be needed to assess the contribution of regional variation and bone ingrowth to implant stability.