Abaloparatide, a novel PTH receptor agonist, increased bone mass and strength in ovariectomized cynomolgus monkeys by increasing bone formation without increasing bone resorption

All animal-related activities were approved by the Charles River Montreal’s Animal Care Committee and performed in an AAALAC-accredited facility. The study was performed under Good Laboratory Practice (GLP) in accordance with the protocol and standard operating procedures at Charles River Laboratories.

Eighty-four female cynos from Mauritius, including spare animals, were received from Primus Bio Resources, aged 9 to 18 years and weighing 3.1 to 6.9 kg. Animals were housed 2–3/stainless steel cage (temperature range 21–27 °C, humidity 30–70%, 12–12-h light-dark cycle), in conformance with accepted AAALAC requirements. Animals had access to standard 2056 Teklad Certified Global Soy-Protein-Free food pellets (Harlan Teklad) and automated watering valves.

A 5-week acclimation period included detailed health exams, including menstrual regularity and body weight. After acclimation, animals underwent DXA scanning. Age, body weight, whole body bone mineral content (BMC), and lumbar spine BMD guided allocation into five treatment groups (n = 15–17/group). Four groups underwent OVX while the fifth group underwent sham surgery, as previously described [15]. After surgery, animals remained untreated for 9 months to allow OVX-induced bone loss. Sham controls and one OVX group then received daily s.c. vehicle (0.9% saline; Sham, n = 15; Veh, n = 17) for 16 months, while other OVX groups received daily s.c. abaloparatide at 0.2, 1, or 5 μg/kg (ABL0.2, ABL1, and ABL5; all n = 16). Dose selection was guided by BMD data from a previous 10-month cyno study, which showed reversal of OVX-induced BMD loss with abaloparatide at 1 and 10 μg/kg/day [17]. Therefore, 1 μg/kg was designated as the pharmacologically effective dose in this species; a fivefold lower dose (0.2 μg/kg) was included as a “suboptimal” dose, and a fivefold higher dose (5 μg/kg) was included to provide an adequate safety margin, in accordance with regulatory guidelines [19, 20].

Regular animal monitoring included food consumption and menstrual status. Detailed weekly clinical evaluations included body weight. At the end of the treatment period, overnight-fasted animals were sedated with i.m. ketamine (15 mg/kg), anesthetized with a weight-based volume of sodium pentobarbital, and euthanized by exsanguination. Gross examinations and full necropsies were performed and internal organs weighed.

Major organs and numerous other tissues were collected at necropsy and evaluated for gross changes. All collected samples were evaluated microscopically for pathologic changes if they exhibited gross changes, and some samples underwent prospective microscopic evaluations, including distal femurs, sternum, 10th thoracic vertebra, heart, kidney, lung, ovary, and uterus.

Dorso-ventral and lateral radiographs of the lumbar and thoracic spine and appendicular long bones (including hip) were taken at the beginning and end of the treatment period and examined for skeletal abnormalities, including any suspected focal proliferative changes that developed or progressed compared with pre-treatment evaluation. Radiographic lesions were collected at scheduled euthanasia, decalcified, and examined microscopically.

Blood was collected by femoral venipuncture after overnight fasting for clinical chemistry (serum calcium, phosphorus, and creatinine; 0.7 mL of blood collected into serum separator tubes) and for bone turnover markers estradiol, PTH(1–84), and 1,25(OH)₂D (6 mL of blood collected into serum separator tubes). Morning urine was collected by catheterization after anesthesia with i.m. glycopyrrolate (0.01 mg/kg), ketamine (5 or 10 mg/kg), xylazine (0.6 mg/kg), or dexmedetomidine (0.01 mg/kg). Serum procollagen 1 N-terminal peptide (P1NP) was measured by UniQ P1NP RIA kits from Orion Diagnostica (Cat. no. 67034), serum bone-specific alkaline phosphatase (BSAP) by MicroVue™ BAP EIA kits from Quidel Corp. (Cat. no. 8012), and serum C-telopeptides (CTX) by Serum CrossLaps ELISA kits (IDS Inc., Cat. no. AC-02F1). Urine N-telopeptides (NTX) was measured by Osteomark NTx Urine ELISA kits (Wampole Laboratories, Cat. no. WL-9006) and adjusted for urine creatinine. Additional serum analyses included estradiol (Ultra-Sensitive Estradiol RIA kit, Beckman Coulter, no. DSL4800), 1,25(OH)₂D (RIA kit, IDS Ltd., no. AA54F2), and PTH(1–84) (human PTH ELISA kit, Immutopics, Inc., no. 60-3100).

Fifteen and 5 days prior to month 16 sacrifice, animals received i.v. bicarbonate-buffered calcein (8 mg/kg). At necropsy, the second lumbar vertebra (L2) and left femur (neck and mid-shaft) were trimmed with a diamond saw and stored in 10% formalin for 3 days. Samples were transferred to 70% ethanol and prepared undecalcified. Tissue blocks were cut through the median plane of the L2 vertebral body and transversely at the left femoral mid-shaft, and the femoral neck was trimmed along the frontal plane. Blocks were dehydrated in ethanol and embedded in methylmethacrylate. Unstained sections from trabecular bone regions were cut at 4–8 μm thickness with a Leica RM2255 rotary microtome and mounted on microscope slides. Unstained sections from cortical bone regions were ground (40–80 μm thick) on an Exakt 400CS micro-grinder and mounted on microscope slides, with two section levels evaluated at the middle of the femoral diaphysis. Trabecular analyses were performed on unstained sections (7 μm) and on sections stained with toluidine blue and Goldner’s trichrome (5 μm). Histomorphometry data were generated by ImagePro Plus® using standard methods and nomenclature [21]. Evaluation of the cancellous bone region was done on the femoral neck (two section levels) and L2 (one section level). Cancellous bone measurements were confined within predetermined lines traced on sections and excluded an approximately 0.25-mm-wide transitional zone between the cancellous and cortical bone tissue. The area of interest (AOI) in L2 comprised two 16-mm² squares traced at each end of the vertebral body, centered through the dorso-ventral plane and starting at 1 mm from the inner edge of the physis. The AOI measured in the proximal femur comprised a 16-mm² square traced in the proximal to mid-femoral neck. Evaluation of cortical bone was performed on two levels of the whole section.

Whenever the mineral apposition rate (MAR) could not be directly measured in a specific bone region of an individual animal due to lack of double fluorochrome labels, a value of 0.2 μm/day was used for MAR of that region in that animal, representing the smallest measurable inter-label distance. For trabecular analyses, the femoral neck of one vehicle control animal exhibited no double label; for cortical analyses, details on labeling status are provided in Section 3.

In vivo DXA was performed with a Hologic Discovery A densitometer; sites and analysis modes included whole body (infant whole body analysis mode), lumbar spine (L1-L4), proximal tibial diaphysis, and right proximal femur (neck plus trochanter, subregion array spine analysis mode). Before each scan, animals were anesthetized by i.m. glycopyrrolate (0.01 mg/kg), ketamine (5 or 10 mg/kg), xylazine (0.6 mg/kg), or dexmedetomidine (0.01 mg/kg). Ex vivo (month 16) DXA scans were obtained for the right femur (total, proximal, central, and distal), L3 and L4 bodies, and L5 and L6 cancellous cores.

In vivo pQCT scans were performed with an XCT Research SA+ bone scanner, software version 5.50. Three sequential scans, 0.5 mm apart, were obtained at the proximal tibia and distal radial metaphysis, with an additional slice in each bone’s diaphysis. The metaphyseal region included total, trabecular, and cortical/subcortical compartments, the latter representing total minus trabecular, while the diaphyseal region comprised cortical bone only. An XCT Research SA bone scanner was used for ex vivo assessments of the right femur, L3-L4 vertebral bodies, and L5-L6 cancellous cores. pQCT scan settings and analysis parameters are provided in Supplemental Table S1. Ex vivo pQCT and DXA data for L3 and L4 bodies were averaged, as were L5 and L6 cancellous core data.

Terminal ex vivo micro-CT was performed with a high-resolution Micro-CT system (Scanco Medical AG micro-CT 100) and 3-D morphometry evaluation program (software version uCT V6.1). The 12th thoracic vertebral body (T12), distal femur, and L6 cancellous cores were scanned at a 34-μm voxel resolution. Humeral beams (described below) were scanned at 5-μm resolution. Supplemental Table S2 provides settings for micro-CT analyses.

Bone samples collected after 16 months of treatment were trimmed and scanned by DXA, pQCT, and/or micro-CT prior to destructive biomechanical testing. L3 and L4 bodies were dissected, leaving parallel-cut caudal and cranial surfaces with exposed trabecular bone. Cylindrical cancellous cores (~ 5 mm diameter) were prepared from the middle region of similarly prepared L5 and L6 bodies. The cores were harvested in the caudal-cranial plane, and each core was then bisected into caudal and cranial halves of approximately 4 mm in height. Trimmed L3 and L4 bodies as well as L5 and L6 cancellous cores, the femoral diaphysis, and the femoral neck were tested using an MTS 858 Mini Bionix Servohydraulic Test System, Model 358.02C. Data were collected using TestSuite TWE version 3.1.2. Cortical beams (1 by 3 by 35 mm) milled from the caudal aspect of the right humeral diaphysis were tested in three-point bending (Bose ElectroForce® 3300 system, WinTest® 4.1). Destructive tests of the right femur (3-point bending), femoral neck (shear test, simulated single-legged stance), vertebral specimens (compression), and humeral beams were conducted and data were calculated as previously described [22], with the following variations: the L3-L4 tests used a 0.5% offset for yield load, and the L5-L6 core tests used a 0.5% offset for yield load and a 2% offset for peak load, stiffness, and work to failure (energy, i.e., area under the load-displacement curve, AUC). L3 and L4 biomechanics data were averaged, as were L5 and L6 cancellous core biomechanics data.

Estimated femoral diaphyseal material properties included ultimate stress (FLc/4I; F = peak load, L = span length, c = radius, and I = cross-sectional moment of inertia), modulus (SL3/48I; S = stiffness), and toughness (0.75 × AUC × b2 / LI; b = diameter). Apparent material properties calculated for L3-L4 bodies and L5-L6 cores included apparent strength (peak load/area), apparent modulus ([stiffness × height] / area), and apparent toughness (AUC / [area × height]).

The success of OVX surgery was confirmed by lack of ovaries at necropsy, reduced uterus weight, and lower serum estradiol compared with sham controls (data not shown). One sham group animal was excluded based on uterine atrophy at necropsy suggesting natural menopause, and another sham animal was euthanized at week 55 due to poor and deteriorating health, potentially related to kidney disease. One Veh animal developed diabetes and died at week 50. One animal from the ABL0.2 group was euthanized on day 7 of treatment due to poor and deteriorating health that was evident during the pre-treatment period. There were no clinical observations related to abaloparatide and no abaloparatide-related changes in food consumption or body weight. Abaloparatide had no adverse effects on gross pathology, organ weights, or histopathology or on clinical chemistry, coagulation, hematology, or urinalysis parameters (data not shown). Radiographic assessments indicated no abaloparatide-related abnormalities that might indicate localized proliferative changes (data not shown).

Abaloparatide had no significant effects on serum levels of calcium, phosphorus, endogenous PTH, 1,25(OH)₂D, or creatinine (Supplemental Table S3) compared with Veh controls.

The bone formation markers serum P1NP and BSAP were significantly higher in Veh versus sham controls at all post-OVX time points, as were the resorption markers serum CTX and urine NTX, consistent with OVX-induced increases in bone remodeling (Supplemental Table S4). Abaloparatide led to greater serum P1NP levels relative to Veh controls, with effects reaching statistical significance at months 3, 7.5, and 12 for the 0.2-μg/kg dose, whereas the two higher doses did not show significance. There were no statistically significant effects of abaloparatide on serum BSAP, serum CTX, or urine NTX at any time point (Supplemental Table S4).

After 25 total months of OVX, including 16 months of treatment, most histomorphometry endpoints for L2, femoral neck, and femoral diaphysis showed no significant differences among groups (Fig. 1 and Supplemental Tables S5, S6, and S7). L2 showed significantly greater trabecular osteoblast surface per bone surface (Tb.Ob.S/BS) and mineralizing surface per bone surface (Tb.MS/BS) in Veh versus sham controls, consistent with OVX-induced increases in bone remodeling (Supplemental Table S5). L2 wall thickness (W.Th) was significantly higher in Veh versus sham controls, and L2 W.Th was significantly higher in all abaloparatide groups versus Veh (Fig. 1a). L2 formation period (FP) was significantly higher in the ABL0.2 and ABL1 groups versus Veh controls (Fig. 1b), driven by increased W.Th without an increase in adjusted apposition rate (Aj.AR) (Supplemental Table S5). L2 osteoclast surface per bone surface (Oc.S/BS) and eroded surface per bone surface (ES/BS) were similar in each abaloparatide group versus Veh controls (Fig. 1c, d).

Trabecular bone parameters at the femoral neck showed few significant differences between groups. Ob.S/BS and W.Th were significantly greater in Veh versus sham controls, and the ABL1 group had significantly greater single-labeled surface per bone surface (sLS/BS) (P < 0.05 versus Veh controls; Supplemental Table S6). Femoral neck Oc.S/BS, ES/BS, and bone formation rate per bone surface (BFR/BS) were similar in abaloparatide versus Veh groups (Supplemental Table S6).

Cortical histomorphometry at the femoral diaphysis indicated that single and/or double Haversian fluorochrome labels were evident in every animal from each group. Likewise, single and/or double endocortical and periosteal labels were evident in all animals in the ABL1 and ABL5 groups. No single or double endocortical labels were observed for seven Sham animals, one Veh animal, and two ABL0.2 animals, and no periosteal labels were observed in seven Sham animals, one Veh animal, and one ABL0.2 animal; for these cases, a mineral apposition rate (MAR) value of 0.2 μm/day was imputed. Significantly greater endocortical labeled surface and endocortical BFR/BS was observed in the femoral diaphysis of Veh controls versus sham controls (Fig. 1e, f). The femoral diaphysis of Veh controls also showed significantly greater Haversian labeled surface (H.L.Pm/H.Pm), higher MAR (H.MAR), and higher BFR (H.BFR/BS) versus sham controls (Supplemental Table S7). The ABL5 group had significantly greater femoral diaphysis endocortical labeled surface and higher endocortical BFR/BS compared with the Veh group (Fig. 1e, f). The abaloparatide groups showed no differences in the resorption parameters endocortical eroded surface or cortical porosity compared with Veh controls (Fig. 1g, h).

In vivo DXA of the whole body and specific axial and appendicular sites indicated that animals from the four combined OVX groups lost significant BMD between the start (month − 9) and end (month 0) of the bone depletion period, as compared with sham controls (Fig. 2). Similar trends in BMD were observed in Sham animals and Veh controls during the treatment period. The ABL1 and ABL5 groups showed significantly greater increases in whole body aBMD, starting as early as treatment month 4 (P < 0.05 versus Veh; Fig. 2a). Each abaloparatide dose significantly increased L1-4 aBMD as early as treatment month 4 (P < 0.05 versus Veh; Fig. 2b). The ABL5 group showed significantly increased cortical aBMC in the proximal tibia at month 16 (Fig. 2c). aBMD of the proximal femur, femoral neck, and trochanter was significantly increased in the ABL1 and ABL5 groups (P < 0.05 versus Veh; Fig. 2d–f).

pQCT indicated that animals from the four combined OVX groups experienced significant bone loss at the proximal tibia during the bone depletion period, as compared with sham controls (Fig. 3). The Veh group exhibited some recovery of tibial bone loss during the treatment period (Fig. 3a–c, g–h). Abaloparatide at 1 and 5 μg/kg fully reversed OVX-induced deficits in proximal tibial total volumetric BMC (vBMC) and vBMD (Fig. 3a, b). All abaloparatide doses also caused significant gains in the cortical/subcortical compartment, including increased cortical/subcortical area and BMD (Fig. 3c, d). pQCT of the proximal tibial diaphysis revealed significant cortical bone gains with abaloparatide: Abaloparatide at 5 μg/kg increased endocortical bone apposition, as shown by significantly lower endocortical circumference versus Veh controls (Fig. 3e), and abaloparatide at 5 μg/kg also increased cortical thickness and vBMC (P < 0.05 versus Veh; Fig. 3g, h). Abaloparatide did not influence cortical vBMD of the proximal tibial diaphysis or the distal radial diaphysis at month 16 (Supplemental Table S8), nor did abaloparatide influence the percentage change in cortical vBMD from month 0 to month 16 at these sites (data not shown). Ex vivo pQCT analyses also suggested that abaloparatide had no significant effects on cortical vBMD of the femoral diaphysis (Supplemental Table S8).

Month 16 micro-CT of the distal femoral metaphysis and the T12 cortical shell indicated lower cortical vBMD in Veh versus sham controls, whereas cortical vBMD in the three abaloparatide groups was similar to Veh controls (Supplemental Table S8). Cortical vBMD of humeral beams was also similar in the abaloparatide and Veh groups (Table S8), as was cortical porosity. Humeral beam cortical porosity as a percentage of cortical volume was 2.30 ± 0.27 (mean ± SEM) for the Sham group, 3.75 ± 0.42 for the Veh group (P < 0.05 versus Sham), 3.85 ± 0.42 for the ABL0.2 group (NS versus Veh), 3.45 ± 0.22 for the ABL1 group (NS versus Veh), and 3.54 ± 0.40 for the ABL5 group (NS versus Veh). Micro-CT of L6 cancellous cores showed lower vBMC and lower bone volume per total volume (BV/TV) in Veh versus sham controls, while the ABL1 and ABL5 groups had greater vBMC and BV/TV compared with Veh (Fig. 4a, b). Abaloparatide also dose-dependently reversed the deleterious OVX-related increase in L6 structural model index (SMI) (Fig. 4c). Representative micro-CT reconstructions of L6 cores for the OVX-Veh and ABL1 groups (Fig. 4d) also depict the denser and more robust trabecular architecture associated with abaloparatide administration.

There were no significant differences between the Veh and other groups for biomechanical parameters of the femoral diaphysis, femoral neck, and humeral beams (Supplemental Tables S9–11). Peak and yield load values for L3-L4 vertebral bodies and L5-L6 cancellous cores were significantly lower in Veh versus sham controls (Fig. 5a–d). The Veh group also had significantly lower values for L3-L4 energy (Supplemental Table S12) and L5-L6 core stiffness (Fig. 5e) compared with sham controls, whereas these parameters were similar in the abaloparatide and Veh groups. The ABL5 group had significantly greater L3-L4 peak load, and the ABL1 and ABL5 groups had significantly greater L5-L6 cancellous core peak load, yield load, stiffness, and energy (all P < 0.05 versus Veh; Fig. 5). Linear regression analyses were performed to assess relationships between bone mass and bone strength, which are considered to represent measures of bone quality. Significant positive relationships were observed for L3-L4 peak load versus QCT-derived total vBMC for all five groups, with r ² values ranging from 0.55 to 0.83. For these regressions, slopes were significantly steeper for the sham, ABL1, and ABL5 groups versus Veh controls, indicating greater L3-L4 strength than might be expected from bone mass (Fig. 5g). For all other bone mass-strength relationships, similar regression lines were observed for each of the five groups, which enabled calculations of overall r ² values for all groups combined. L5-L6 core peak load correlated positively with ex vivo QCT vBMC (Fig. 5h), with an overall r ² of 0.87 (P < 0.05). Femoral shaft peak load correlated positively with femoral shaft cortical vBMC, with an overall r ² of 0.85, and femoral neck peak load correlated positively with femoral neck aBMC, with an overall r ² of 0.54 (both P < 0.05) (Fig. 6).