Cathepsin K (CatK) is a lysosomal cysteine proteinase that is highly expressed in osteoclasts (OC) and plays a key role in bone resorption.1 During bone resorption, CatK is secreted under the ruffled border and accumulates in the acidified resorption lacunae where it degrades matrix proteins. It is activated at low pH, allowing the breakdown of collagen molecules generating N- and C-telopeptides, in addition to triple helical region fragments.2 We and others have shown that knockout mice for CatK develop osteopetrosis because of decreased bone resorption by osteoclasts.3, 4 Elimination of CatK does not appear to have any other nonskeletal developmental consequences. Osteoclasts in CatK knockout (CatK−/−) mice express increased levels of Trap-5b, cathepsin L, and matrix metalloproteases 9 and 14,5 which may represent a compensatory phenomenon owing to the loss of collagenolytic activity. Adult CatK−/− mice show higher BMD at the central and distal femur and thicker cortices than wild-type mice.3 In contrast, transgenic mice overexpressing CatK display low bone mass and an increased bone turnover rate.5 Pycnodysostosis, a rare autosomal recessive skeletal disorder characterized by high bone mineral density (BMD), acroosteolysis of the distal phalanges, short stature, and skull deformities, is linked to several loss-of-function mutations in the CatK gene.6–8
In preclinical models, the traditional rat and mouse models of bone loss are not appropriate for testing in vivo efficacy of CatK inhibitors because of low homology between the rodent and human CatK active sites. Several CatK inhibitors with limited potency toward the rodent enzymes prevent trabecular bone loss in ovariectomized (OVX) rats or mice at high doses.9–11 Because of the high molecular identity of rabbit to human CatK enzymes, we previously studied the ability of human CatK inhibitors to prevent estrogen-deficiency bone loss in rabbits.12 However, the nonhuman primate (NHP) and human CatK enzymes are also highly homologous, and OVX-NHP has long been considered the standard model for postmenopausal osteoporosis when evaluation of Haversian and trabecular bone remodeling is the goal.13
Relacatib is a potent human CatK inhibitor with a limited selectivity profile against other cathepsins. In estrogen-deficient skeletally mature female and young intact male cynomolgus monkeys, relacatib treatment resulted in a reduction in bone turnover with improved BMD and bone strength.14 In a 9-month dose ranging study comparing relacatib with alendronate (ALN) in OVX cynomolgus monkeys,15 relacatib preserved lumbar spine areal BMD (aBMD) and total femur neck volumetric BMD (vBMD). Dynamic histomorphometry demonstrated that relacatib reduced bone formation and resorption at trabecular sites in the same manner as ALN. Although intracortical remodeling was suppressed by ALN, it was preserved at low doses of relacatib.15
Odanacatib (ODN, MK-0822) is a highly selective, potent, and reversible inhibitor of human CatK.16 It has ∼300-fold selectivity against cathepsin S and ≥1000-fold selectivity against all other human cathepsins.16 Enzyme kinetic studies have demonstrated that ODN behaves as a fully reversible time-dependent inhibitor. ODN was previously reported to inhibit osteoclast-mediated bone resorption in vitro.16 ODN dose-dependently blocked bone loss in adult newly OVX rabbits with efficacy similar to ALN.12 However, unlike ALN, ODN had no effect on bone formation rate in the vertebrae and femur.3 ODN exhibits a half-life of ∼18 hours in rhesus monkeys and is therefore suitable for daily dosing in NHPs.16
ODN is currently being developed as an orally bioavailable therapeutic for the treatment of postmenopausal osteoporosis. In a phase I study, ODN demonstrated an approximate half-life of 50 to 60 hours, suitable for weekly dosing.17 In a phase II clinical trial, once-weekly ODN treatment dose-dependently increased lumbar spine and total-hip BMD and suppressed bone turnover markers in postmenopausal women with low bone mineral density (BMD).18 The purpose of this study was to evaluate the effect of ODN at two dose levels on BMD, bone turnover markers, and biomechanical properties at the lumbar spine of skeletally mature OVX rhesus monkeys. In addition, the effects of ODN on histomorphometric parameters of bone resorption, bone formation, and osteoclast morphology in transilial biopsies at 12 months and in the trabecular region of the lumbar vertebral body after 21 months of treatment were also evaluated.
Materials and Methods
Rhesus monkeys, diets, surgery, and dosing with ODN
The in-life portion of this study was conducted at the New Iberia Research Center (New Iberia, LA, USA) with the approval of both the University of Louisiana at Lafayette and Merck Institutional Animal Care and Use Committees (IACUCs). Forty-two female rhesus monkeys (Macaca mulatta, aged 13 to 23 years) were recruited into this study as retired breeders, occasionally cycling, however, without childbearing for at least 3 years. All subjects received a physical examination, including complete blood count and serum chemistry screens before study. Examination of radiographs confirmed epiphyseal closure in the long bones. Upon arrival at the New Iberia Research Center, the animals were acclimated for a minimum of 1 month before surgery and placed on a high-protein diet (19.8%) containing 1.17% Ca, 0.7% P, 8 IU/g vitamin D3 (Harlan-Teklad [Indianapolis, IN, USA] 8773 NIB primate diet; ∼200 g/d) for the duration of the study. The monkeys were individually housed and provided with drinking water and daily enrichment ad lib and weighed monthly. Before OVX surgery, all monkeys were randomized into groups by weight and baseline lumbar spine (L1 to L4) bone mineral density (LsBMD).
All monkeys were fasted 12 hours before surgery, then preanesthetized with ketamine HCl (Teva Animal Health, St. Joseph, MO, USA) (10 mg/kg, intramuscularly [im]). Cefazolin was administered (20 mg/kg, im) 1 hour presurgery and again at 2 hours postsurgery. General anesthesia was maintained using isoflurane at 1% to 1.5% throughout the surgical procedure. Bilateral OVX surgery was performed by ligation of the utero-ovarian proper ligament and blood vessels. Each ovary was excised with as much surrounding tissue as possible. Postoperative care was performed according to the IACUC recommendation. Nalbuphine (0.05 to 0.1 mg/kg, im) was given postoperatively for pain management and then every 4 hours for the first 12 hours. Additional nalbuphine (7 mg/kg, t.i.d.) was given on an as-needed basis. According to IACUC recommendation, no sham surgery was performed to avoid surgery in NHPs that produces no physiological alterations.
Intact animals (n = 11) were dosed daily with vehicle (0.5% methocel plus 0.1% SDS; Sigma-Aldrich; St. Louis, MO, USA). The p.o. OVX animals were divided into three groups: vehicle (n = 11), ODN, 6 mg/kg p.o., q.d. (n = 10), or ODN 30 mg/kg p.o., q.d. (n = 10). ODN was dosed using vehicle, and treatment started at ∼11 days postsurgery and continued for 21 months. The animals were dosed daily via orogastric intubation using a dosing volume adjusted monthly according to individual body weight. Pharmacokinetic (PK) studies were done at 30, 92, and 149 days after the initial dosing to determine drug exposure levels. Plasma samples were collected at 0, 1, 2, 4, 7, and 24 hours postdosing. A 12-hour plasma sample was also collected at day 92.
Transilial and rib biopsy and necropsy
After 12 months of treatment, calcein (Sigma; 8 mg/kg, subcutaneously [sc]) was given to all monkeys 18 and 3 days before transilial and rib biopsy. Transilial biopsies (TIBx) were performed while the animal was in lateral recumbency. Anesthesia was as for ovariectomy. A 10 × 10-cm region over the iliac crest was shaved, and a 5- to 6-cm skin incision over the craniodorsal aspect of the iliac crest was made. A Rochester bone trephine (8 mm internal diameter) was used to perform a biopsy at a site ∼10 mm below the anterolateral aspect of the iliac crest. The trephine was carefully advanced laterally through both cortices, while rotating, then removed. Finally, the cylindrical transilial biopsy specimen was gently extruded from the trephine and fixed in 70% ethanol.
For the rib biopsy (RBx), a 5 × 5-cm region over the seventh rib area, 4 to 6 cm from the costo-chondral junction, was shaved, scrubbed, and prepared for skin incision as previously described.19 The periosteum was incised. A pair of cuts separated by 2 cm were made with a Gigli wire saw. Sharp edges on the remaining portion of the rib were dulled by a bone file, and routine closure of the surgical site was performed. Pain management and incision site management as for OVX was provided. The 2-cm biopsy specimen was immediately placed in 70% ethanol.
The intact group (n = 11) was recruited to another study and thus no ex vivo data from the intact group will be presented. One monkey from the intact group was euthanized because of severe pulmonary edema after 16 months on study; one monkey from the ODN 6 mg/kg group was euthanized because of significant lung pathology, abscesses, and adhesions after 10 months on study; and one monkey from the ODN 6 mg/kg group developed pneumothorax, post RBx at 12 months. Individual animals were dosed with tetracycline (250 mg/kg, p.o.) at 18 and 3 days before necropsy. The left femora and lumbar vertebrae (L1 to L2) were fixed in 70% ethanol and stored at room temperature for histomorphometric analyses. The rate of bone turnover in rib, transilial biopsies, and lumbar spine trabecular bone were analyzed by histomorphometry. The right femora and lumbar vertebrae (L4 to L5) were wrapped in saline-soaked gauze and stored at −20°C until analyzed for bone strength.
Bone turnover markers
Serum and 24-hour urine were collected at baseline (time 0, before surgery), 1.5, 3, 6, 9, 12, 15, 18, and 21 months postsurgery and stored at −80°C until analyzed for markers of bone resorption and formation. Biomarkers of bone resorption included serum carboxyl-terminal telopeptide of type 1 collagen (sCTx) determined using CrossLaps ELISA (IDS Ltd., Fountain Hills, AZ, USA); urinary amino-terminal telopeptide of type 1 collagen (uNTx) measured with Osteomark NTx ELISA (Inverness Medical, Princeton, NJ, USA); and total urinary deoxypyridinoline (uDPD) determined using the total DPD ELISA (Quidel, Santa Clara, CA, USA). All urinary markers were normalized to creatinine, which was determined by spectrophotometric assay (TECO Diagnostics, Anaheim, CA, USA). Serum markers of bone formation including N-propeptide of type 1 collagen (P1NP) and bone-specific alkaline phosphatase (BSAP) were determined using assays from IDS Ltd.
Other biomarkers measured include serum carboxyl-terminal telopeptide of type 1 collagen generated by metalloproteinases (1-CTP) using an ELISA assay (IDS Ltd.) that recognizes the trivalently linked C-telopeptide fragment of type 1 collagen. This fragment is produced through the action of matrix metalloproteinases rather than the physiological action of CatK. The osteoclast-specific marker, serum tartrate-resistant acid phosphatase 5b (Trap-5b) was also determined using Bone TRAP assay (IDS Ltd.), a solid-phase immunofixed enzyme assay.
In vivo dual-energy X-ray absorptiometry (DXA)
The animals were anesthetized by ketamine and maintained under isoflurane as for OVX. Areal BMD, bone mineral content (BMC), and bone area of the lumbar vertebrae L1 to L4 were measured by DXA using a Hologic QDR Discovery A bone densitometer (Hologic, Inc., Waltham, MA, USA). Other skeletal sites including whole body, total hip, distal radius, and 1/3 distal radius were also scanned. In vivo DXA scans of animals were scheduled at baseline (before surgery for randomization) and approximately at 3, 6, 9, 12, 16, and 20 months after study start. Ex vivo DXA was performed on the excised L4 to L5 vertebrae.
Ex vivo peripheral quantitative computed tomography (pQCT)
pQCT was used to measure BMC and BMD of cross sections taken at the midsection of the vertebrae excised for biomechanical testing, L4 and L5. Single-slice pQCT scans were acquired (voxel size 0.20 mm; contour mode 2; threshold 0.446 1/cm; peel mode 20; 60% trabecular area) using a XCT Research SA or SA + bone scanner with software version 5.50D (Stratec Medizintechnik, Pforzheim, Germany). Volumetric measures of total and trabecular BMD and BMC were obtained from a 0.8-mm-thick slice taken from the middle of each vertebral body.
Histology and histomorphometric analyses
For trabecular bone histomorphometry, three sites were analyzed: TIBx and RBx taken at 12 months and lumbar vertebra (L2) at 21 months. L2 was cleaned of muscle and most connective tissue, then spinous and transverse processes were removed. A sagittal cut was made through the body of the vertebra and the right portion embedded undecalcified in 80% methyl methacrylate (Polysciences, Warrington, PA, USA). Six-micron parasagittal sections of L2 were cut on a Leica SM 2500S sledge microtome (Leica Biosystems, Heidelberg, Germany) and mounted on glass slides (Fisher Scientific, Pittsburgh, PA, USA). Coverslips were applied using Eukitt's mounting media (EMS, Ft. Washington, PA, USA). TIBx and RBx were processed similarly.
For L2, measurements were made in a 3 × 4-mm area of the 1/3 caudal aspect of each section ∼2 mm from the caudal growth plate and centered between the dorsal and ventral sides. Unstained sections were analyzed for tetracycline labeling. Osteoclast morphology was examined in sections stained by a modified toluidine blue method.20
Because of trabecular bone scarcity in some TIBx and RBx samples, both the endocortical (Ec) and trabecular (Tb) bone components were measured in three section planes, each 100 µm apart with the first plane starting at 2 mm from the edge of the TIBx. The RBx was sectioned longitudinally to maximize trabecular bone. The final values were derived from the sum of measurements in all three planes and provided a measurable perimeter of ≥15 mm for Ec surfaces and 19 to 36 mm for Tb perimeter. Histomorphometric analysis was performed separately in each Ec or Tb compartment. Goldner trichrome-stained sections in TIBx were used to measure osteoid and toluidine blue-stained sections or Von Kossa/Trichrome-stained sections were used to measure eroded surface and osteoclast surface in both TIBx and RBx. For TIBx and RBx, measurements of calcein label were completed on unstained sections. All measurements were made using a light/epifluorescent microscope, Nikon Eclipse 80i (Nikon Instruments, Melville, NY, USA) interfaced to an image analysis system (Bioquant R&M Biometrics, Nashville, TN, USA).
Histomorphometric variables were measured, calculated, and abbreviated as previously described.21 Dynamic histomorphometric endpoints measured were mineralizing surface (MS/BS, %), mineral apposition rate (MAR, µm/d) and bone formation rate (BFR/BS, µm3/µm2/year). Mineralizing surface was calculated based on double-labeled perimeter plus half single-labeled perimeter divided by total bone perimeter. From double-label width and interlabel time period, mineral apposition rate (MAR, µm/d) was calculated. BFR/BS was calculated by multiplying MS/BS and MAR. Bone specimens without double label were excluded from the calculation of mean MAR but were assigned a BFR of zero.22
Static morphometric endpoints measured include: osteoid surface (OS/BS, %), osteoid thickness (O.Th, µms) (TIBx only), osteoclast surface (Oc.S/BS), osteoclast number (N.Oc/BS), and eroded surface (ES/BS) (TIBx and RBx only). In L2, additional measurements were made to include osteoblast surface (Ob.S/BS), osteoclast number (N.Oc/BS) with or without toluidine blue-stained vesicles per bone surface, total osteoclast number per bone surface, and the number of nuclei per osteoclast (# Nc/Oc).
Bones were thawed overnight in a refrigerator at 4°C before ex vivo DXA, pQCT, and bone strength testing. Vertebrae were trimmed, removing dorsal elements, spinal processes, and the cartilaginous end plates with a diamond saw, leaving only the cartilage-free vertebral body with plano-parallel ends. Biomechanical testing was performed using the 858 Mini Bionix Servohydraulic Test System, Model 242.03.22 All data were collected using Excel Spreadsheet, which was used to manually derive the required biomechanics parameters. The height of the vertebral body test specimen was measured using digital calipers before compression testing with an MTS 858 Mini Bionix with a load cell of 15 kN and a loading rate of 20 mm/minute. Peak load was recorded as the maximum of the load-displacement curve, and stiffness was the slope of the linear portion. Work to failure (AUC), the energy required to break the bone, was calculated as the area under the curve to peak load for compression tests. The apparent strength was calculated using the peak load divided by the area. Modulus was calculated using stiffness, thickness, and area. Toughness was calculated using AUC, thickness, and area.13, 23, 24 Data were not collected for one L4 of one animal in the ODN 6 mg/kg group because of a technical oversight. Only L5 data were used for the calculations of this animal.
All statistical comparisons of areal BMD, bone turnover markers, and histomorphometric parameters were analyzed using Statview software (SAS Institute Inc., Cary, NC, USA). All data in tables and figures are shown as means ± standard error (SEM). Differences among three or more groups were analyzed by one-way analysis of variance (ANOVA), followed by Fisher's least significant difference (Fisher's PLSD), except where indicated. Differences were considered significant when p ≤ 0.05.
For the statistical analyses of biomechanical testing and pQCT data, group variances were compared using Levene's test. When differences between group variances were not significant, a one-way ANOVA was performed. When significant, Dunnett's test determined differences between groups. Whenever Levene's test indicated heterogenous group variances (p ≤ 0.05), the nonparametric Kruskal-Wallis test was used to compare groups. If the Kruskal-Wallis test was significant, then the significance of the differences between the control group and each test group were assessed using Dunn's test. Correlation analysis of ex vivo densitometry-derived parameters and biomechanical strength parameters was performed using a Pearson correlation test.
The baseline characteristics of the animals are shown in Table 1. The monkeys were randomized first with respect to lumbar spine aBMD (L1 to L4) and then to body weight. The groups were matched with respect to age, serum clinical chemistry, and whole body BMD (Table 1). Also note that the baseline values of the bone turnover markers, uNTx, sCTx, P1NP, and BSAP, among the groups were not different (Fig. 1).
Table 1. Baseline Characteristics of Monkeys
Intact + vehicle
OVX + vehicle
OVX + ODN (6 mg/kg)
OVX + ODN (30 mg/kg)
ANOVA p value
Mean ± SEM.
Body weight (kg)
7.82 ± 0.43
7.79 ± 0.39
7.82 ± 0.37
7.83 ± 0.43
9.47 ± 0.11
9.60 ± 0.15
9.48 ± 0.07
9.34 ± 0.09
4.45 ± 0.04
3.49 ± 0.19
4.02 ± 0.34
3.71 ± 0.36
L1 to L4 DXA
0.68 ± 0.02
0.67 ± 0.02
0.67 ± 0.03
0.67 ± 0.03
12.51 ± 0.60
12.51 ± 0.26
12.42 ± 0.51
12.79 ± 0.54
18.37 ± 0.58
18.71 ± 0.57
18.52 ± 0.52
18.98 ± 0.22
0.33 ± 0.01
0.31 ± 0.01
0.32 ± 0.01
0.32 ± 0.01
214 ± 6
198 ± 7
209 ± 10
203 ± 8
Total hip DXA
0.69 ± 0.01
0.63 ± 0.02
0.63 ± 0.02
0.67 ± 0.03
5.34 ± 0.18
4.53 ± 0.09
4.69 ± 0.25
4.94 ± 0.25
At baseline, body weights of the monkeys ranged from 5.89 to 11.13 kg, and averaged 7.886 ± 0.204 kg (mean ± SEM). Body weights were increased by 8.1% to 9.9% in all groups during the treatment phase. Three animals were excluded from the analyses for reasons described above and determined to be nondrug-related. Pharmacokinetic studies were performed at ∼1, 3, and 5 months after the initial dosing. ODN at a dose of 6 mg/kg resulted in an exposure of 2.2 µM/24 hours. ODN at 30 mg/kg provided an exposure of 4.2 µM/24 hours.
Effect of ODN on bone resorption and bone formation markers
In the OVX-vehicle group, increases in bone resorption markers were detected by 6 weeks post-OVX. Urinary NTx levels (Fig. 1A) in OVX-vehicle animals significantly reached a plateau at ∼140% at 3 months post-OVX compared with baseline. The levels remained elevated up to a year and gradually declined to 50% to 80% above baseline by the end of the study (Fig. 1A). A similar modest response to OVX was observed for sCTx (Fig. 1B) and uDPD (data not shown). Bone formation markers responded more slowly to OVX than resorption markers, reaching significance by 3 months postsurgery compared with intact (Fig. 1C,D). P1NP significantly increased to ∼250% over baseline at month 3 in response to OVX, indicating rapid bone turnover. Levels remained elevated for more than a year and gradually declined to ∼80% over baseline toward the end of the study (Fig. 1C). BSAP significantly increased by ∼50% over baseline by month 3 and declined to baseline at 18 months of treatment (Fig. 1D).
From the absolute changes in response to ODN treatment versus baseline values (Fig. 1), we calculated percent change relative to OVX-vehicle. ODN at 6 mg/kg reduced uNTx by 75% and further reduced uNTx levels by 90% at 30 mg/kg relative to OVX-vehicle throughout the study duration. ODN-dependent reduction of uNTx remained significantly below OVX-vehicle at all time points (Fig. 1A). Similarly, ODN dose-dependently decreased sCTx (Fig. 1B). ODN 6 mg/kg reduced sCTx by ∼40% compared with OVX-vehicle, restoring levels to that of intact animals (Fig. 1B). ODN 30 mg/kg significantly reduced sCTX by ∼55% compared with OVX to levels below those in intact animals. ODN trended to reduce uDPD by 10% to 20% with the 6 mg/kg dose and by 40% with the 30 mg/kg dose versus OVX-vehicle. Although the levels of uDPD were highly variable, ODN clearly restored levels to or below those of intact animals at all time points after 3 months of dosing (data not shown). The results demonstrated that ODN effectively decreased uDPD, a collagen degradation marker known to be independent of CatK enzymatic activity.
After 3 months, P1NP and BSAP significantly peaked at 246% and 56%, respectively, above baseline in the OVX-vehicle animals. ODN at both doses consistently reduced P1NP levels throughout treatment to levels observed in intact animals (Fig. 1C). Although initially BSAP levels trended upward in the 6 mg/kg group, at 3 months, treatment with ODN at both the 6 mg/kg or 30 mg/kg doses decreased BSAP by 30% and 35%, respectively, compared with OVX-vehicle.
Effects of ODN on 1-CTP and TRAP-5b
As a C-terminal peptide product of collagen I generated by metalloproteinases and further degraded by CatK to generate CTx, 1-CTP is normally undetectable in postmenopausal women and not responsive to bisphosphonates.25 Similarly, 1-CTP did not respond to OVX in this study (Fig. 2A). However, as a direct substrate of CatK, this marker was significantly elevated in ODN-treated monkeys. From the absolute changes relative to baseline, levels of 1-CTP were calculated to increase by ∼50% in the ODN 6 mg/kg group compared with intact and OVX from 6 to 21 months and by 50% to 100% in the ODN 30 mg/kg group at all time points compared with intact and OVX-vehicle monkeys.
Osteoclast-specific Trap-5b is an indicator of osteoclast number.26 In this study, Trap-5b trended to increase in response to OVX by 20% at 3 months and significantly increased by 90% to 110% above baseline levels until the study termination (Fig. 2B). Neither dose of ODN had an effect on Trap-5b levels throughout the experiment.
Effect of ODN on BMD of the lumbar spine
Lumbar spine bone mineral density (LsBMD) at L1 to L4 in the intact group was 0.677 ± 0.020 g/cm2 at baseline and 0.752 ± 0.029 g/cm2 after 20 months of dosing, a significant increase of 11% versus baseline (Fig. 3A,B). LsBMD in the OVX-vehicle group was 0.673 ± 0.018 g/cm2 at baseline and was reduced to 0.665 ± 0.021 g/cm2 at 21 months of treatment, a 1.8% decrease versus baseline (Fig. 3A,B).
Treatment with ODN 6 mg/kg increased LsBMD in OVX monkeys from 0.714 ± 0.017 g/cm2 at baseline to 0.764 ± 0.021 g/cm2 at 20 months, an increase of 7.2% from baseline (Fig. 3A), significantly higher than that in the OVX-vehicle group (p < 0.05) and not different from intact (Fig. 3A,B). Note that after excluding the three monkeys that died, the ODN 6 mg/kg group had higher BMD at baseline but was not statistically different from other groups. This led to the apparent higher mean aBMD in the ODN 6 mg/kg group compared with intact and vehicle-treated OVX monkeys in the cross-sectional analysis at 20 months (Fig. 3B). Treatment with ODN 30 mg/kg increased LsBMD in OVX monkeys from 0.692 ± 0.021 g/cm2 at baseline to 0.784 ± 0.032 g/cm2 at 20 months, an increase of 15% versus baseline (p < 0.05 versus OVX-vehicle and versus ODN 6 mg/kg) as shown in Fig. 3A,B.
Effect of ODN on bone strength parameters at the lumbar spine
Consistent trends in bone densitometry and bone strength parameters were noted and are discussed below. However, compared with OVX-vehicle controls, differences for bone strength endpoints did not attain statistical significance (Table 2). The biomechanical parameters (peak load, stiffness, toughness, apparent strength, area under the curve [AUC] and modulus) trended to be higher (up to 36%) for ODN-treated animals compared with OVX-vehicle controls (Table 2). The trended increases in biomechanical parameters were consistent with the dose-dependent increases in QCT BMC and BMD (up to 7% and 12%, with 6 and 30 mg/kg, respectively) compared with OVX-vehicle controls. For all groups combined, a statistically significant positive linear correlation was obtained for peak load versus QCT-derived BMC (R = 0.838, p < 0.0001; Fig. 4) as well as peak load versus QCT-derived BMD (R = 0.852, p < 0.0001; data not shown).
Table 2. QCT Analysis and Biomechanical Testing of Lumbar Vertebrae (Averaged L4 and L5) of OVX Rhesus Monkeys Treated With ODN for 21 Months
OVX + vehicle
OVX + ODN (6 mg/kg)
OVX + ODN (30 mg/kg)
Mean ± SEM.
L4 to L5 (n)
421.6 ± 18.8
450.8 ± 21.5
470.3 ± 29.1
65.6 ± 3.0
68.2 ± 3.4
72.9 ± 5.2
Peak load (N)
3528 ± 239
4120 ± 331
4338 ± 382
Apparent strength (MPa)
22.9 ± 1.8
27.4 ± 2.2
28.0 ± 2.3
0.9 ± 0.2
1.1 ± 0.2
1.2 ± 0.2
15,221 ± 1053
15,110 ± 1379
17,523 ± 1614
826.3 ± 63.6
858.8 ± 88.1
951.2 ± 83.2
1234.6 ± 198.1
1378.0 ± 243.1
1591.7 ± 327.2
Effect of ODN on bone formation in transilial biopsy and lumbar vertebrae
Transilial biopsies (12-month treatment)
Values for bone formation parameters in TIBx Ec and Tb surfaces are shown in Fig. 5B. In trabecular bone, bone formation rate (BFR/BS) in OVX animals was 100% higher (p < 0.01) than in intact animals. Compared with OVX-vehicle, though, ODN 6 mg/kg significantly decreased Tb.MAR by about 40% (p < 0.05); ODN 30 mg/kg had no significant effect on Tb.MAR (Fig. 5B-b). Ec.MAR was not affected by ODN (Fig. 5B-e). ODN treatment at both doses reduced Tb.MS/BS and Tb.BFR/BS by 70% to 80% compared with OVX-vehicle (p < 0.01; Fig. 5B-a,c). Similarly, ODN at both doses decreased Ec.MS/BS and Ec.BFR/BS by 60% to 80% compared with OVX-vehicle (p < 0.05; Fig. 5B-d,f). The ODN doses did not differ in their effects on BFR/BS and MS/BS. Bone formation endpoints were reduced significantly by ODN in both Ec and Tb compartments of the TIBx. Reduction of bone formation by ODN in the rib biopsies (RBx) was similar to that in the TIBx (data not shown).
Lumbar vertebrae (21-month treatment)
The effects of OVX and treatment with ODN on trabecular MS/BS, MAR, and BFR/BS at the L2 are shown in Fig. 6. In this study, intact animals were not euthanized and recruited to another study. Therefore, dynamic histomorphometric measurements of the treatment effects of ODN on L2 are compared with the vehicle-treated OVX group. ODN at 30 mg/kg decreased MS/BS by 68% (p < 0.05 versus OVX-vehicle). No significant ODN-related effects on MAR were observed. Hence, BFR/BS was decreased by 30 mg/kg ODN (p < 0.05) compared with the vehicle-treated OVX group.
Effects of ODN on eroded surface, osteoclast surface, and number and morphology in transilial biopsies and lumbar spine
Transilial biopsies (12-month treatment)
Treatment with ODN at either dose induced accumulation of darkly toluidine blue-stained intracellular vesicles in all osteoclasts. No stained vesicles were ever observed in osteoclasts from OVX-vehicle or intact controls. Micrographs of representative toluidine blue-stained osteoclasts are shown in Fig. 7A.
Osteoclast surface (Oc.S/BS), osteoclast number (N.Oc./BS), and eroded surface (ES/BS) values for both Tb and Ec compartments of TIBx are summarized in Table 3 and Fig. 7B. ODN had no significant effect on Oc.S/BS and N.Oc/BS compared with OVX-vehicle. However, compared with intact animals, Oc.S/BS (Table 3) and N.Oc./BS (Fig. 7B-a,b) were significantly higher in the ODN 6 mg/kg group (p < 0.05). Results for RBx are similar (data not shown).
Table 3. Histomorphometry of Transilial Biopsies (12-Month Treatment) and Lumbar Spine (21-Month Treatment)
Intact + vehicle
OVX + vehicle
OVX + ODN (6 mg/kg)
OVX + ODN (30 mg/kg)
Mean ± SEM.
Different from intact and OVX + vehicle, p < 0.05.
Eroded surface (ES/BS) at endocortical and trabecular regions was not significantly affected by OVX. ES/BS at the Tb surface was significantly higher with 6 mg/kg ODN than in the OVX-vehicle group (p < 0.05) (Table 3).
Lumbar spine (21-month treatment)
In the trabecular bone of L2, compared with OVX-vehicle, 30 mg/kg ODN increased both N.Oc./BS and OcS/BS fourfold (p < 0.01; Table 3 and Fig. 8B). Nuclei number per osteoclast (#Nc/Oc) and distribution of the relative frequencies of #Nc/Oc were unchanged by ODN treatment (Fig. 8C,D).
No toluidine blue-stained vesicles were detected in osteoclasts of the OVX-vehicle group. In the ODN groups, 98% and 96% of the osteoclasts in the 6 mg/kg and 30 mg/kg groups, respectively, had stained vesicles (Fig. 8A-a). Toluidine blue-stained vesicles in osteoclasts appeared to be a marker of the pharmacologic action of ODN.
Effects of ODN on osteoid and osteoblasts in transilial biopsies and lumbar spine
Transilial biopsy (12-month treatment)
Osteoid surface (OS/BS) and osteoid thickness (O.Th) values within the Tb and Ec compartments of TIBx are summarized in Table 3. In both compartments, OVX increased OS/BS and O.Th by 50% to 70% with all differences significant (p < 0.01) except for Tb.OS/BS. ODN at both 6 mg/kg and 30 mg/kg doses prevented these increases with no dose dependency (p < 0.01).
Lumbar spine (21-month treatment)
Treatment with 30 mg/kg ODN significantly reduced Tb.Ob.S/BS by 83% versus OVX-vehicle (p < 0.001) (Table 3).
Cathepsin K has become a high-profile target for the development of selective inhibitors for the treatment of diseases associated with high bone turnover, including osteoporosis.27, 28 Estrogen deficiency in NHPs is a model of osteoporosis in humans that has been used frequently to study drug effects in an environment that mimics postmenopausal bone loss.13, 29–31 The primary objective of this study was to assess the effects of the CatK inhibitor ODN on bone turnover and BMD in trabecular bone sites in newly ovariectomized adult rhesus monkeys. A secondary endpoint was the study of bone strength in the lumbar vertebral body. The efficacy of ODN on cortical bone sites in these OVX monkeys will be reported elsewhere.
In this study, lumbar spine aBMD was significantly lower in vehicle-treated OVX monkeys than in the intact group at 20 months postsurgery. Likely the result of changes in diet, longitudinal monitoring indicated that there was bone gain in the intact animals and a failure to lose bone mass in OVX animals. The monkeys in this study were skeletally mature breeders from domestically reared colonies, who had borne no offspring for at least 3 years. Upon arrival at the New Iberia Research Center, the animals were switched to a high-protein, calcium/phosphate, and vitamin D3 diet. The change in diet may explain the observed gain in BMD in the spine of intact monkeys and very mild bone loss in the OVX monkeys. Changes in diet and physical activity in OVX-NHP have also been previously reported to prevent estrogen-induced bone loss.31 Visual examination at necropsy and ex vivo micro-CT imaging at necropsy were performed to detect the presence of spinal osteophytes. Approximately 10% of the animals had osteophytes in lumbar vertebrae (L1 to L2) and were evenly distributed among the groups (data not shown). It is not known whether osteophytosis progressed during the 21 months of study duration because no prestudy imaging of osteophytosis was performed.
Although there was no significant loss in lumbar spine BMD in the OVX monkeys, bone turnover markers increased significantly in response to OVX. This increase quickly reached a plateau and gradually declined after 15 months to near baseline levels of intact monkeys. The study subjects began a daily dosing regimen with either vehicle or ODN at 6 mg/kg (2.2 µM/24 hours) or 30 mg/kg (4.2 µM/24 hours) 11 days after OVX. The two doses of ODN resulted in lower daily exposures than the estimated daily exposure in patients receiving the 50-mg weekly dose (6 to 7 µM/24 hours) selected for the phase III clinical trial. Nonetheless, both doses of ODN effectively reduced high bone turnover in the newly OVX monkeys and significantly increased lumbar spine BMD of these animals compared with the vehicle-treated controls.
The relationship of peak load to bone mineral content was positive and highly correlated, indicating that normal bone biomechanical properties were maintained by ODN treatment during the 21 months of dosing. Although the study groups were randomized, the sizes of the monkeys recruited into the study ranged broadly at baseline. In addition, a relatively small group size (n = 8 to 11/group) was used for biomechanical strength testing at the end of the study. Together, the low study power for this cross-sectional analysis may be responsible for the apparent lack of significant treatment effects on bone strength endpoints. Hence, we combined data from all groups for regression analysis and concluded that although fully effective in increasing bone mass in OVX monkeys to equivalent levels as in the intact group, ODN maintained normal bone biomechanical properties post-treatment for 21 months. We also note that treatment effects of ODN on microarchitecture and strength of the distal radius from the ODN-treated OVX monkeys compared with vehicle control were also quantified using high-resolution peripheral QCT (or Xtreme-CT) and compression loading configuration finite element analysis (FEA).32, 33 Measurable differences were observed in the geometry, average volumetric BMD, and FEA estimated ultra distal radial bone strength between the ODN-treated groups compared with vehicle-treated estrogen-deficient nonhuman primates. The results from this study will be reported elsewhere.
Although relacatib is a potent inhibitor of human CatK (Ki = 41 pM), it is not selective against CatL and CatV and exhibits limited selectivity (39- to 300-fold) against other cathepsins.14 A recent study described OVX cynomolgus monkeys treated daily for 9 months with relacatib at 1, 3, and 10 mg/kg (p.o., q.d.) compared with alendronate (ALN) at 0.05 mg/kg, iv once every 2 weeks.15 Only the higher dose of relacatib (10 mg/kg) prevented spine and hip BMD loss similar to ALN. The magnitude and duration of the effects of this dose of relacatib on resorption markers and formation markers, such as BSAP, were generally comparable to that of ALN. Our findings with ODN in OVX NHPs appear very similar to earlier studies of relacatib and balicatib in NHPs. However, relacatib had no effect on the levels of osteocalcin in OVX-cynomolgus monkeys. Indeed, in this study, ODN also had no effect on osteocalcin in OVX monkeys (data not shown), suggesting that this biomarker may not reflect osteoblastic bone formation in response to therapy with a CatK inhibitor.
The doses of ODN were selected according to the relative exposure, which resulted in suppression of bone resorption markers (uNTx) in NHP (2 to 4 µM/24 hours) in a pilot study (data not shown). In this study, both doses (6 and 30 mg/kg) of ODN blocked OVX-induced relative lumbar spine bone loss and reduced all bone resorption markers (sCTx, uNTx, and uDPD) compared with OVX animals. In addition, the levels of uNTx were reduced by ODN to below intact levels throughout the dosing period. ODN at both doses also decreased bone formation markers (P1NP and BSAP) to intact levels. These findings in OVX monkeys with lower exposures of ODN and daily dosing were different from those with ODN 50 mg weekly in a 24-month phase II study with postmenopausal women with osteoporosis.18 In the clinical study, ODN treatment modestly and transiently decreased P1NP and BSAP levels, whereas in this monkey study, ODN treatment persistently reduced these bone formation markers to levels similar to those in monkeys treated with a bisphosphonate.29 The peak daily exposure level of the CatK inhibitors achieved in monkeys versus the weekly peak exposure of ODN in postmenopausal women may explain the differential effects of this drug on the bone formation markers. Moreover, the mechanism of CatK inhibition on bone turnover reduction may also be greatly influenced by the high remodeling rate in the newly ovariectomized NHP, compared with the relatively lower remodeling state in postmenopausal women with osteoporosis.
In addition to the catalytic activity of CatK on type 1 collagen degradation, several other matrix metalloproteinases (MMP) and cathepsins are known to participate in the initiation of the bone resorption process and are coupled to bone formation.34 At the C-telopeptide terminus of type 1 collagen, two fragments have been characterized: 1) the cross-linked carboxyterminal telopeptide (1-CTP) and 2) the linear eight amino acid sequence of C-terminal cross-linked telopeptide (CTx). CTx has been shown by in vitro testing to be efficiently generated by CatK cleaving the 1-CTP epitope.35 Here, we demonstrated an increase in serum 1-CTP after 6 weeks of treatment with ODN. The 1-CTP level reached steady state by ∼3 to 6 months with both doses of ODN in OVX monkeys. The steady state level of 1-CTP persisted throughout the study, suggesting that there was no continuous upregulation of MMP or other cathepsin resulting in the release of 1-CTP in order to compensate for ODN-induced long-term reduction in CTx levels. It should be noted that this marker is not normally measured in osteoporotic patients or in preclinical models of estrogen deficiency.36 Moreover, bisphosphonate therapy had no effects on 1-CTP levels in preclinical or clinical studies.25 Previously, we reported that ODN increased 1-CTP in osteoporotic patients.17 Here, we showed dose-dependently elevated levels of 1-CTP in OVX rhesus monkeys, suggesting that this epitope may be a sensitive target engagement biomarker for CatK in response to ODN therapy.
Trap-5b has been classified as a resorption marker, but recent observations have suggested it may serve as a biomarker for osteoclast number.26 Cathepsins K and L have been reported to efficiently cleave and activate Trap-5b.36–38 However, proteolytic cleavage of Trap-5b was not abolished in CatK−/− mice.38 Here, histologic analyses demonstrated that compared with OVX-vehicle, osteoclast numbers trend to increase in the lumbar spine of the monkeys treated with ODN for 21 months. In addition, ODN maintained the elevated levels of Trap-5b associated with OVX, suggesting inhibition of CatK did not decrease osteoclast number. Taken together, 1-CTP and Trap-5b are readily used as convenient markers to differentiate the mechanism of CatK inhibition from antiresorptives including bisphosphonates and anti-RANKL therapies. Our findings in the OVX monkeys are also in accordance with previous studies showing that osteoclast number was either unaffected or tended to increase in CatK knockout mice.3 An increased ratio of RANKL to OPG was suggested to be responsible for this increase of osteoclast number.39 Impaired osteoclastic bone resorption in CatK−/− mice results in activation of osteoblastic cells to produce increased amounts of RANKL in vivo.22, 39 Here, pharmacological inhibition of CatK activity by ODN may also mediate osteoclast-derived signals, leading to locally increased RANKL expression and enhanced osteoclastogenesis. Alternatively, enhanced diurnal parathyroid hormone secretion after administration of once-daily doses of a CatK inhibitor, balicatib, for 14 days in Japanese postmenopausal women was suggested to contribute not only to inhibition of bone resorption but also to stimulation of bone formation.40 This feedback mechanism may also be responsible for the ODN-mediated increase in osteoclastogenesis.
We recently demonstrated that trabecular bone formation was significantly higher in CatK knockout mice than in age- and gender-matched wild-type mice.3 We have also shown in OVX rabbits that ODN prevented bone loss while sparing bone formation at multiple skeletal sites.12 Although limited numbers of biopsies were collected in the dose-ranging phase II study, no apparent suppression of bone formation was detected in iliac crest biopsies of patients receiving ODN.18 Presently, it is unclear why the bone formation sparing action of CatK inhibition in ODN-treated OVX rabbits was not observed in the trabecular bone of the newly OVX monkeys treated with the same CatK inhibitor. The rabbit has been recognized as an excellent model to study the repair of bony defects in tissue engineering models. The use of rabbit bone marrow stromal cells has been proven to possess strong osteogenic potential.41 Hence, the species differences in bone marrow osteogenic potential may explain in part the above discrepancy. On the other hand, histomorphometric evidence in femoral cortical bone sites demonstrated that ODN spared endocortical bone formation in the femur and dose dependently stimulated periosteal bone formation.42, 43 These findings suggest a bone site-specific action for ODN in bone formation in monkeys, consistent with previous reports with other CatK inhibitors balicatib and relacatib in OVX NHP.15, 44 Moreover, the impact of ODN on bone resorption and formation markers may be more representative of the overall effects of the inhibition of high bone turnover rate at the more numerous trabecular surfaces and less at slow bone turnover cortical sites.
Fibroblasts and osteoclasts from patients with pycnodysostosis accumulate nondigested collagen fibers in endocytic vesicles.34 However, osteoclast morphology in these patients was not significantly altered on Goldner-stained sections.45 Osteoclasts from patients treated for approximately 1 year with balicatib (AAE581) have been reported to have increased numbers of intracellular vesicles enriched for Trap-5b.46 However, these cells did not exhibit an apoptotic morphology and were distinct from the hypermultinucleated giant osteoclasts seen in patients after long-term treatment with bisphosphonates.47 Similar morphological changes in osteoclasts, as seen in balicatib-treated patients, were also observed in the current NHP study. The cells had intracellular vesicles stained darkly with toluidine blue. ODN-treated osteoclasts otherwise appeared to be healthy, retaining the same distribution of nuclei without changes in size or displaying apoptotic appearance. Currently, we are conducting studies to further characterize the contents of these vesicles and the reversibility of this morphologic change. Everts and colleagues34 also reported that periosteal fibroblasts from pycnodysostosis patients had 18-fold higher collagen-containing vacuoles than fibroblasts from control patients. In this study, toluidine blue-stained vesicular accumulation was detected only in osteoclasts and no accumulation was observed in other cell types in the periosteal or bone marrow compartments. Furthermore, large areas of demineralized bone were reported to be accumulated on bone surfaces of pycnodysostotic patients.34, 45 Conversely, here we demonstrate that ODN treatment reduced osteoid surface and thickness at 12 months in transilial biopsies and at 21 months in lumbar spine.
In summary, this study assessed the long-term effects of the CatK inhibitor ODN on bone efficacy and safety, utilizing several endpoints of bone mass, turnover, histology, and strength in the OVX rhesus monkey, a relevant animal model for human osteoporosis. The results indicate that ODN inhibits bone resorption as assessed by biomarkers and prevents estrogen deficiency–induced bone loss in the lumbar spine. The early and sustained changes in bone resorption markers were accompanied by later changes in bone mass and trabecular histomorphometry, in which ODN reduced trabecular bone resorption and bone formation endpoints. Furthermore, long-term treatment with ODN maintained normal bone biomechanical properties in the lumbar spine. This therapy also led to characteristic increased serum 1-CTP levels and maintained OVX-induced Trap-5b levels. Taken together, our findings demonstrate that the CatK inhibitor ODN inhibits bone resorption and estrogen deficiency–induced bone loss in nonhuman primates, via a molecular mechanism distinct from the bisphosphonates.
PJM, BLP, MP, KRS, GAW, BBS, and LTD are employees of Merck, the study's sponsor, and may own stock/stock options in the company. DBK was an employee of Merck during the conduct of this study. SYS, RS, and JEG have no disclosures to declare.
We thank the excellent veterinary and technical staff in the New Iberia Research Center. Kenneth Lodge and Cory Merschman assisted with specimen collection. We also thank Judy Pun and Charles Chen for their assistance in the analyses of the rib and transilial biopsies.
Authors' roles: Study design: PJM, BLP, MP, SYS, RS, BBS, DBK, LTD. Study conduct: PJM, BLP, MP, KRS, GAW, RS, JEG. Data collection: PJM, BLP, MP, KRS, GAW, RS, JEG. Data analysis: PJM, BLP, MP, SYS, RS, JEG, DBK, LTD. Data interpretation: PJM, BLP, MP, KRS, GAW, SYS, RS, JEG, BBS, DBK, LTD. Drafting manuscript: PJM, BLP, MP, KRS, BBS, LTD. Revising manuscript content: PJM, BLP, MP, KRS, GAW, SYS, RS, JEG, BBS, DBK, LTD. Approving final version of manuscript: PJM, BLP, MP, KRS, GAW, SYS, RS, JEG, BBS, DBK, LTD. LTD takes responsibility for the integrity of the data analysis.