Recombinant Human Parathyroid Hormone (1–34) [Teriparatide] Improves Both Cortical and Cancellous Bone Structure

Authors

  • Yebin Jiang MD, PhD,

    Corresponding author
    1. Osteoporosis and Arthritis Research Group, Department of Radiology, University of California, San Francisco, California, USA
    • Osteoporosis and Arthritis Research Group Department of Radiology University of California San Francisco, CA 94143-0628, USA
    Search for more papers by this author
    • Dr Jiang received a research grant from Eli Lilly and Company for this study. Dr Genant is on the speaking bureau at Eli Lilly and Company. Drs Mitlak and Wang are employees and shareholders of Eli Lilly and Company. Dr Eriksen is an employee of Eli Lilly and Company.

  • Jenny J Zhao,

    1. Osteoporosis and Arthritis Research Group, Department of Radiology, University of California, San Francisco, California, USA
    Search for more papers by this author
  • Bruce H Mitlak,

    1. Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, Indiana, USA
    Search for more papers by this author
  • Ouhong Wang,

    1. Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, Indiana, USA
    Search for more papers by this author
  • Harry K Genant,

    1. Osteoporosis and Arthritis Research Group, Department of Radiology, University of California, San Francisco, California, USA
    Search for more papers by this author
  • Erik F Eriksen

    1. Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, Indiana, USA
    Search for more papers by this author

Abstract

Histomorphometry and μCT of 51 paired iliac crest biopsy specimens from women treated with teriparatide revealed significant increases in cancellous bone volume, cancellous bone connectivity density, cancellous bone plate-like structure, and cortical thickness, and a reduction in marrow star volume.

Introduction: We studied the ability of teriparatide (rDNA origin) injection [rhPTH(1–34), TPTD] to improve both cancellous and cortical bone in a subset of women enrolled in the Fracture Prevention Trial of postmenopausal women with osteoporosis after a mean treatment time of 19 months. This is the first report of a biopsy study after treatment with teriparatide having a sufficient number of paired biopsy samples to provide quantitative structural data.

Methods: Fifty-one paired iliac crest bone biopsy specimens (placebo [n = 19], 20 μg teriparatide [n = 18], and 40 μg teriparatide [n = 14]) were analyzed using both two-dimensional (2D) histomorphometry and three-dimensional (3D) microcomputed tomography (μCT). Data for both teriparatide treatment groups were pooled for analysis.

Results and Conclusions: By 2D histomorphometric analyses, teriparatide significantly increased cancellous bone volume (median percent change: teriparatide, 14%; placebo, −24%; p = 0.001) and reduced marrow star volume (teriparatide, −16%; placebo, 112%; p = 0.004). Teriparatide administration was not associated with osteomalacia or woven bone, and there were no significant changes in mineral appositional rate or wall thickness. By 3D cancellous and cortical bone structural analyses, teriparatide significantly decreased the cancellous structure model index (teriparatide, −12%; placebo, 7%; p = 0.025), increased cancellous connectivity density (teriparatide, 19%; placebo, −14%; p = 0.034), and increased cortical thickness (teriparatide, 22%; placebo, 3%; p = 0.012). These data show that teriparatide treatment of postmenopausal women with osteoporosis significantly increased cancellous bone volume and connectivity, improved trabecular morphology with a shift toward a more plate-like structure, and increased cortical bone thickness. These changes in cancellous and cortical bone morphology should improve biomechanical competence and are consistent with the substantially reduced incidences of vertebral and nonvertebral fractures during administration of teriparatide.

INTRODUCTION

Once-daily administration of parathyroid hormone (PTH), the major hormonal regulator of calcium homeostasis, causes increased bone formation and bone mass.(1) After the first report in 1929 of increased skeletal calcium in rats after injection of parathyroid extract,(2) preclinical studies and small clinical trials have shown pronounced anabolic effects of intermittent PTH administration on bone.(3,4)

Most recently, a large randomized double-blind multicenter study, the Fracture Prevention Trial tested recombinant human PTH(1–34) [teriparatide, rhPTH(1–34), TPTD] versus placebo for treatment of osteoporosis in 1637 postmenopausal women.(5) Daily injections of 20 or 40 μg of teriparatide over a mean of 19 months increased bone mineral density (BMD) at the lumbar spine and proximal femur and significantly decreased the incidence of vertebral and nonvertebral fractures.(5)

The effect of injected PTH on human cortical bone is controversial. Some small early clinical studies found that appendicular BMD was reduced by PTH treatment, whereas vertebral BMD increased.(6–8) In other studies, BMD at the predominantly distal radius or femoral neck(9) changed little during PTH administration.(10–12) These findings led to speculation that the anabolic effects of teriparatide on cancellous bone may be obtained at the expense of cortical bone.

In most studies of PTH peptides to treat osteoporosis, the primary endpoint was a change in BMD. Only four such studies have examined the effects of PTH peptides on iliac crest bone histology.(13–16) However, those were small, used different protocols, and had variable results. Three studies(13,14,17) found that PTH peptides significantly increase cancellous bone volume, and the other two studies found significant increases in cortical thickness by two-dimensional (2D)(15) and three-dimensional (3D)(16) analyses.

Stereological analysis of iliac crest bone structure based on 2D sections can potentially yield 3D structural information. However, the amount of tissue available for analysis is small, because only a few thin sections of the biopsy core are used. This leads to pronounced variation of calculated structural indices and various possible biases.(18,19) Analysis of 3D structure of bone biopsy specimens is now possible using a variety of X-ray- or nuclear magnetic resonance (NMR)-based techniques, among which the most widely used is microcomputed tomography (μCT).(19) These techniques permit measurements of the entire biopsy core and therefore may provide more stable and unbiased estimates of bone structural indices.

This study presents quantitative microstructural data from a substudy of the Fracture Prevention Trial, in which a subset of patients randomized to placebo or teriparatide injections underwent iliac crest biopsies before and after teriparatide administration. Based on the antifracture efficacy observed in these patients, it was hypothesized that teriparatide treatment would improve both cancellous and cortical bone in postmenopausal women with osteoporosis. Changes in bone remodeling and trabecular and cortical microstructure in paired iliac crest bone biopsy specimens were analyzed using both 2D histomorphometry(20) and 3D μCT.(19)

MATERIALS AND METHODS

Subjects

One hundred and two patients participated in the biopsy study, and a total of 51 paired iliac crest biopsy specimens, of sufficient quality for analysis, were obtained from subjects at 11 sites in five countries participating in the randomized, multicenter, double-blind, placebo-controlled Fracture Prevention Trial. BMD and fracture incidence were the primary clinical outcomes of this trial. The characteristics of these patients and primary outcomes of the study have been previously described in detail.(5) Briefly, women were eligible for enrollment if a period of at least 5 years had elapsed since menopause, and if they had at least one moderate or two mild atraumatic vertebral fractures on radiographs of the thoracic and lumbar spine, and an ambulatory status. For women with fewer than two moderate fractures, an additional criterion for enrollment was a value for BMD of the hip or lumbar spine that was at least 1 SD below the mean value in normal premenopausal white women (age range, 20–35 years). BMD was measured by DXA. Women were excluded from the study if they had illnesses that affect bone or calcium metabolism.

Patients signed informed consent to the treatment and investigation protocol, which was approved by the Institutional Review Board for Research Involving Human Subjects, at each participating center.

Treatment

Teriparatide was synthesized at Lilly Research Laboratories using recombinant DNA technology, lyophilized, and stored at −20°C. All enrolled subjects were supplemented with a daily intake of 1000 mg of elemental calcium and 400–1200 IU of vitamin D3. The women administered a once-daily self-injection of placebo for 2 weeks and were then randomly assigned to receive placebo or 20 or 40 μg of teriparatide daily. Bone mass was measured by DXA of the lumbar spine, total hip region, and forearm.

Specimens

Transiliac crest bone biopsy specimens were obtained from 102 women, enrolled in the Fracture Prevention Trial, using a Bordier needle trephine system (8 mm inner diameter) after in vivo double labeling with tetracycline fluorochromes given orally in a 3:12:3 day sequence. All sites performing biopsy specimens were instructed to use sites 2 cm behind and 2 cm below the anterior superior iliac spine. They were also instructed to insure that the biopsy was taken perpendicular to the axis of the ilium in that area. The first biopsy was taken at baseline and the second from the contralateral iliac crest after 12 months of treatment or at treatment endpoint. Treatment duration for these women at the time of the second biopsy was 18 ± 5 months (range, 11–24 months). The biopsy specimens were fixed in 70% ethanol, dehydrated in a graded series of alcohols (70–100%, two changes per grade, each for 4 h under vacuum), and placed in xylene and infiltrated with methylmethacrylate under vacuum at 20 psi on a 2 h/step and 24 h infiltration cycle. The samples were then embedded in methyl methacrylate and sectioned on a heavy-duty microtome (Jung K2). Fifty-one paired iliac crest biopsy specimens (placebo [n = 19], 20 μg teriparatide [n = 18], and 40 μg teriparatide [n = 14]) were found to be complete and of sufficient quality to be analyzed by both histomorphometry and μCT. The criterion to reject a biopsy sample was the loss of one cortex.

Histomorphometry

Biopsy specimens were measured without the knowledge of sample treatment group. All static parameters of remodeling activity were measured under light microscopy on 5-μm-thick biopsy sections stained with Goldner's trichrome. Dynamic parameters of bone formation were also measured under fluorescent microscopy of unstained 15-μm-thick biopsy sections at 125× magnification.(20) Point counting and intersection-based assessments were used to measure most static and dynamic parameters, and a digitizer was used to measure the distance between tetracycline labels and osteoid thickness. All measurements represent the mean of four sections with 100-μm spacing. Measured and derived variables were expressed according to the standard nomenclature and formulas recommended by the American Society of Bone and Mineral Research nomenclature committee.(18)

The static histomorphometric parameters for trabecular bone included total bone volume per tissue volume (BV/TV, %), fraction of trabecular surface covered by osteoid (OS/BS, %), and total erosion surface (ES/BS, %). Mean wall thickness (W.Th, μm) of completed osteons was derived from measurements between the cement line and quiescent bone surface. Erosion depth (E.De, μm) was assessed by lamellar counting as described by Eriksen.(20) Trabecular diameter (Tb.Dm, μm), trabecular number (Tb.N, mm−1), and trabecular separation (Tb.Sp, μm) were calculated indirectly from the bone surface to volume ratio and BV/TV. Marrow star volume (Ma.St.V, mm3), an index of trabecular connectivity, was measured at a 15-fold magnification using a projection microscope and computed as the sum of the lengths of rays projected from randomly selected points in the marrow space toward trabeculae.

Static histomorphometric parameters for cortical bone were measured for the inner and outer cortex and averaged for each biopsy. Cortical tissue was analyzed between the periosteal and endosteal surfaces. The junction between endosteal and trabecular bone was selected to exclude endocortical trabeculae. Cortical thickness (Ct.Th, μm) was expressed as the average of grid intersection-based, evenly spaced, orthogonal intercepts across each of the two cortices. Cortical porosity (Ct.Po/Ct.V, %) was expressed as the average percent area occupied by Haversian canals within the inner and outer cortical tissue and was assessed by point counting.

Both double and single tetracycline-labeled surfaces (dLS/BS, %; sLS/BS, %) were measured to obtain the mineralizing surface (MS/BS, %), using the correction for “label escape” (dLS + 0.5 × sLS). Dynamic parameters also included mineral apposition rate (MAR, μm/day).

μCT imaging acquisition

Non-decalcified specimens embedded in methyl methacrylate were trimmed to fit the specimen holder and examined using a compact fan-beam-type μCT scanner (Scanco Medical AG, Bassersdorf, Switzerland), which can work in either a spiral scanning or multislice mode. A small X-ray tube with a 10-μm nominal diameter μfocal spot was used as a source. The detector consisted of a linear charge-coupled device (CCD) array. A scout view scan was obtained for selection of the examination volume of the specimens, by automatic positioning, measurement, and offline reconstruction.(21)

For each sample, a total of approximately 600 microtomographic sections were acquired with a slice increment of 17 μm. The field of view was 17 × 17 mm2, and the matrix size was 1024 × 1024. Images with isotropic resolution of 17 μm3 were obtained.

μCT imaging processing and analysis

μCT image acquisition and analysis of the biopsy specimens were performed without the knowledge of sample treatment group. For subsequent image analyses, subvolumes of the library of 600 sections originally obtained were selected. Each selected volume of interest (VOI) contained either trabecular bone or cortical bone. To exclude endocortical trabeculae, the trabeculae within 0.5 mm of the endocortical surface were excluded as the junction between endosteal and trabecular bone. To avoid the risk of subvolume selection bias, precautions were taken to exclude a buffer zone at the junction between endosteal and trabecular bone consisting of endocortical trabeculae of 0.5 mm width starting from endocortical surface. The volumetric data were recorded in binary data.

3D trabecular structural parameters were measured directly, as previously described.(19) Because the X-ray attenuation throughout a non-homogeneous material is not uniform and there may exist trabeculae of varying densities throughout the specimen, the selection of one commonly used global and local gray-scaled threshold value, or constraint region growing segmentation method, to create a binary image was inappropriate.(22,23) Mineralized bone was separated from bone marrow with a 3D segmentation algorithm based on the analysis of the steepest gradient calculated from a continuous polynomial fit least-squares approximation of the originally discrete CT volume to find digital edges, which were most commonly characterized by intensity changes in a local image neighborhood. For example, smooth transitions between neighboring intensities are called roof edges, representing points of change from increasing to decreasing intensities or vice versa, whereas a sudden change between neighboring intensities is termed step edge, which describes the borderline between two adjacent regions with considerably different intensities.(23) Bone surface area was determined using the matching cubes method to triangulate the surface of the mineralized bone phase using an interpolating 3D surface reconstruction algorithm.(23) The marching cube algorithm decides how a cube made of neighboring voxels is intersected by the original rugged surface. The resulting surface is a polygon represented by triangles. A voxel within a cube can either belong to the object or to the background. Various triangulated surfaces result from the intersection of the surface and the cube, depending on the voxel configurations with a cube. The algorithm marches on to the next cube, after detecting the surface of the investigated cube in the discrete data set. This approach smooths the surface, which has a 3D polygonal representation consisting exclusively of triangles.

BV was calculated using tetrahedrons corresponding to the enclosed volume of the triangulated surface. TV was the volume of the sample that was examined. A normalized index, BV/TV, was used to compare samples of varying size. The methods used for calculating trabecular thickness (Tb.Th), Tb.Sp, Tb.N, and structural model index (SMI) have been described previously.(19)

In aging and postmenopausal osteoporosis, the deterioration of trabecular bone structure is characterized by a change from plate elements to rod elements. SMI quantifies the characteristic form of a 3D structure in terms of the amount of plates and rods composing the structure. The SMI is based on a differential analysis of the triangulated bone surface. The SMI value is 0 and 3 for an ideal plate structure and a perfect cylindrical rod structure, respectively, independent of the physical dimensions. For a structure with both plates and rods of equal thickness the value lies between 0 and 3, depending on the volume ratio of rods and plates. Morphological measures such as the Euler number, a measure of the maximum number of branches that could be removed before the structure was divided into multiple pieces, was determined in the μCT data set without prior skeletonization. The connectivity (C) of a two-component system, i.e., bone and marrow, was derived directly from the Euler number, by C = 1 − E, if all the trabeculae and bone marrow cavities are connected without isolated marrow cavities inside the bone.(24) It was normalized by examined tissue volume and reported as connectivity density (CD).

Ct.Th was expressed as the average thickness of the 3D cortex on the inner and outer cortical surfaces of the biopsy specimens. A semiautomatic contouring algorithm defined the cortical margins with a manual tracing correction locating periosteal and endosteal surfaces or the most probable edge of the inner and outer cortical surfaces. Filling maximal spheres in the whole cortex with the distance transformation, then calculating the average thickness of all bone voxels determined Ct.Th. Cortical porosity (Ct.Po) was the nonbone in the binary image and was calculated with the same procedure as for BV/TV.

To determine reproducibility of the μCT examination, 20 specimens from different groups were rescanned and reanalyzed. The root mean square CV of the measurements was 2.6% (BV/TV), 3.6% (Tb.N), 5.9% (Tb.Th), 4.0% (Tb.Sp), 3.3% (DA), 2.1% (SMI), 3.9% (CD), 2.7% (Ct.Po), and 2.9% (Ct.Th).

Statistical analysis

Recently, the antifracture efficacy of teriparatide was reported,(5) and treatment with either 20 or 40 μg/day of teriparatide was shown to have similar antifracture efficacy. In addition, this biopsy substudy did not reveal any significant difference in structural variables between the two teriparatide treatment groups. Therefore, because of the small sample size, the data for both doses of teriparatide were pooled for the analyses of 2D and 3D structural parameters. Patient demographics were compared across treatment groups and with the total population of the Fracture Prevention Trial, using the Pearson χ2 test for discrete variables and t-test for continuous variables. For each measured variable, individual patient's paired baseline and endpoint data were analyzed to construct percent change from baseline values, and these data were summarized for both the placebo group and the pooled teriparatide group. The 2D and 3D biopsy data were not normally distributed; therefore, rank transformation was applied before statistical analyses were performed. The placebo and teriparatide treatment groups were compared using t-test (based on ranked or unranked data), and the summary statistics for biopsy specimens were expressed as median and interquartile range (25th, 75th) for each individual group.

RESULTS

The baseline demographic (Table 1) and bone structural characteristics (Tables 2 and 3) of study participants were similar across all study groups. Baseline demographics for this substudy population did not differ from those of the total Fracture Prevention Trial population.(5) Pairwise comparisons between 20 and 40 μg teriparatide treatment groups showed no significant p values for 2D (p > 0.05) or 3D (p > 0.1) structural variables. Therefore, the data for the teriparatide treatment groups were pooled for all analyses.

Table Table 1. Baseline Characteristics of a Subset of Women From the Fracture Prevention Trial Who Provided Paired Biopsies
original image
Table Table 2. Percent Change From Baseline of 2D Trabecular and Cortical Bone Histomorphometry Values
original image
Table Table 3. Percent Change From Baseline in 3D Trabecular and Cortical Bone Microstructure
original image

BMD

Baseline and endpoint measurements of BMD were obtained for all biopsy study participants, and data for the 51 patients with complete paired biopsy specimens are shown in Table 4. Data from patients with incomplete biopsy specimens were not significantly different from patients with paired biopsy specimens. There was a significant increase in BMD for the teriparatide treatment group compared with placebo at the lumbar spine (teriparatide, +14.4% ± 1.3%; placebo, +0.3% ± 1.6%; p < 0.001), femoral neck (teriparatide, +6.3 ± 0.9; placebo, −2.4% ± 0.9; p < 0.001), and intertrocanter (teriparatide, +6.1 ± 0.9; placebo, −1.9% ± 0.7; p < 0.001). Measurements of BMD at the radius shaft and distal radius were not significantly different between placebo and teriparatide treatment groups.

Table Table 4. Percent Change in BMD From Baseline for Placebo and Pooled Teriparatide Treatment Groups*
original image

2D histomorphometry

Qualitative analysis of the biopsy specimens (Fig. 1) revealed no osteomalacia or woven bone in the teriparatide groups. Increased trabecular and cortical thickness was also observed in most post-treatment biopsy specimens. However, in biopsy specimens obtained from women treated with 40 μg teriparatide, marrow fibrosis (2/14) and tunneling resorption (4/14) were seen. Changes in histomorphometric indices reflecting activity and turnover (ES/BS, OS/BS, MS/BS, MAR, and Ac.F) did not differ between groups. Neither did changes in erosion depth (E.De) or wall thickness (W.Th; Table 2).

Figure FIG. 1..

2D sections of two sets of paired iliac crest bone biopsy specimens. Improved trabecular connectivity as well as increased cortical thickness after teriparatide treatment is seen in both sections. A significant increase in BV/TV and decrease in Ma.St.V were observed after treatment with teriparatide (p < 0.05). A and C are baseline samples, and B and D are samples after treatment with (A and B) 20 μg teriparatide or (C and D) 40 μg teriparatide.

Analysis of structural indices (Table 2) revealed a significant increase in cancellous bone volume in the pooled teriparatide group versus placebo (teriparatide, 14% [0%, 50%] [median percent change, 25th, 75th interquartile range]; placebo, −24% [−48%, 8%]; p = 0.001). Teriparatide also caused reductions in marrow star volume (teriparatide, 16% [−42%, 47%]; placebo, 112% [11%, 212%]; p = 0.004). In the placebo group, cortical thickness decreased, whereas it increased in the pooled teriparatide group. However, because of the pronounced variation in this index, this trend did not reach significance. No significant increase in cortical porosity was observed in the pooled teriparatide-treated group.

3D μCT

A representative picture derived from the μCT analysis of one paired teriparatide biopsy sample is shown in Fig. 2. In the post-treatment biopsy, an improvement of the osteoporotic trabecular structure is seen compared with the pretreatment biopsy, with a change toward more plate-like morphology and an increase in cortical thickness. The percentage decrease in SMI (teriparatide, −12% [−43%, 11%] [median % change, 25th, 75th interquartile range]; placebo, 7% [−14%, 39%]), increase in CD (teriparatide, 19% [−15%, 84%]; placebo, −14% [−32%, 32%]), and increase in Ct.Th (teriparatide, 22% [7%, 41%]; placebo 3% [−19%, 13%]) were found to be significant (p < 0.05) after teriparatide treatment, based on the calculation of percentage change from each individual patient (Table 3). However, no significant increase in 3D trabecular bone volume fraction was demonstratable (teriparatide, 7% [−15%, 99%]; placebo, −5% [−40%, 38%]; p = 0.098; Table 3). There was no statistically significant difference in cortical porosity between placebo and pooled teriparatide-treated patients.

Figure FIG. 2..

3D μCT reconstructions of a paired iliac crest bone biopsy. The cortical bone of the iliac crest and trabecular microstructure and connectivity between the cortex can be observed in these images. Compared with the (A) baseline biopsy, treatment with (B) 20 μg of teriparatide increased trabecular bone volume, trabecular connectivity, and cortical thickness. Note also the change in trabecular morphology from a rod-like structure to a more plate-like pattern in the post-treatment biopsy. This paired biopsy sample was obtained from a 65-year-old woman who was treated with 20 μg teriparatide for 21 months.

DISCUSSION

In this study, both 2D histomorphometry and 3D μCT techniques were used to analyze biopsy specimens from a subset of postmenopausal women with osteoporosis enrolled in the Fracture Prevention Trial. This approach provided quantitative data to assess changes in bone structural indices and clearly showed that teriparatide treatment was able to improve both cancellous and cortical bone structure. This is the first report of quantitative microstructural data from paired biopsy specimens obtained from patients treated with teriparatide.

In contrast to a large increase in BMD observed in these patients after treatment with teriparatide,(5) this biopsy study revealed no significant changes in cancellous bone remodeling indices, bone resorption indices (erosion surface, erosion depth, and resorption period), bone formation rates, or labeling indices after 18 months of teriparatide treatment. From previous short-term studies, where biopsy specimens were obtained after 1 month,(16) we know that teriparatide is able to increase labeled surface and increase bone formation very rapidly on quiescent surfaces. Lindsay et al. has reported that biochemical markers of bone resorption and formation reached a maximum level after 1–6 months of teriparatide treatment, which gradually returned to baseline over the duration of the study.(12) The lack of increase in bone turnover in our study is probably explained by the late timing of the biopsy specimens in relation to initiation of therapy (11–24 months). At this time, bone turnover may have been reduced toward baseline levels. We were also unable to show significant increases in wall thickness, albeit a positive trend was seen. It may be that early in the treatment cycle, cancellous packets of increased thickness may have been laid down and were then subsequently covered with packets of more normal thickness by the time of the second biopsy. Dempster et al. previously reported an increased width of endocortical walls but also found the width of trabecular packets to be unchanged.(16) The lack of change in trabecular wall thickness in both studies may be explained by several factors. First, wall thickness was assessed over the bone surface as a whole, which means that a significant proportion of walls measured were formed before teriparatide treatment was initiated. Second, a large proportion of new bone formation in the early phases of teriparatide therapy may occur on quiescent surfaces, not previously resorbed.(16,25) This may contribute to the improvement in trabecular structural indices but not necessarily an increase in wall thickness.

Previously, PTH, like other anabolic agents such as sodium fluoride,(26) was thought to exert its effect mainly on cancellous bone. Moreover, there are unresolved questions that PTH might exert deleterious effects on cortical bone in terms of cortical thinning(6,8,13,), 27 because of an increase in intracortical and endocortical resorption.(28) Our study shows that teriparatide actually improved cortical bone structure, reflected in a 22% increase in cortical thickness. The images obtained during the μCT analysis suggest that the increased cortical thickness resulted from increased bone formation at both the periosteal and endosteal surfaces. Potentially, the early increases in bone turnover seen during intermittent teriparatide therapy could exert negative effects on cortical bone strength through an increase in porosity. However, in rabbits, although intracortical remodeling was significantly increased by teriparatide treatment, the subsequent increase in intracortical porosity did not compromise the mechanical strength of the bone.(29) It seemed that the increased porosity occurred primarily at the endocortex, which has been proven to have a lesser effect on the mechanical properties of bone than if this transient loss of bone were to occur at the periosteal surface.(24) Similar observations of conserved bone strength despite increased cortical porosity have also been reported after teriparatide treatment in a cynomolgus monkey model.(30) Finally, the clinical experience with teriparatide during the Fracture Intervention Trial found a tendency for nonvertebral fractures to be reduced after the ninth month of the study.(5)

Discrepancies between the histomorphometric and μCT analysis for indices of cortical thickness and cancellous bone volume may be explained by amount of tissue examined using the two different techniques and by the segmentation threshold based on mineralization density that is used in μCT imaging. The newly formed bone on the cancellous bone surface may not be optimally mineralized, and therefore not as dense as the segmentation threshold, and unable to be detected by μCT. However this new, relatively hypomineralized bone could be detected by histological staining. On the other hand, the mineralization of the cortical bone was much denser than the cancellous bone, and even the newly formed cortical bone could be detected by μCT. Failure of histology to detect the change in cortical thickness could be caused by the limited amount of tissue examined.

Studies in human bone reported by Dempster et al.(16) showed that an increase in cortical bone was accompanied by increased endocortical wall width and reduced erosion perimeter, suggesting a positive balance at the endosteal envelope. Our in vivo assessment of bone structure using peripheral quantitative computed tomography (pQCT)(31) and radiogrammetry(32) also demonstrated improved cortical structure in two different subgroups of women in the Fracture Prevention Trial. Zanchetta et al.(31) demonstrated increased diameter but unchanged cortical thickness of the radius after treatment with teriparatide using pQCT measurements, leading to significant increases in axial and polar moments of inertia. Furthermore, using radiogrammetry, Hyldstrup et al. (32) demonstrated increased cortical thickness of metacarpals and the radius and reduced marrow diameter, suggesting both increased endosteal and periosteal bone formation. Finally, a structural analysis of hip geometry by Uusi-Rasi et al.,(33) based on DXA scans from the Fracture Prevention Trial, also revealed significant increases in cortical thickness. Increased cortical thickness contributes substantially to bone strength even at skeletal sites such as the vertebrae, where cancellous bone has traditionally been thought to play the dominant role in determining strength.(34,35) One study reported that cortical bone is responsible for up to two-thirds of the mechanical strength of vertebrae in osteoporotic patients.(35)

Teriparatide also improved trabecular architecture and increased trabecular bone volume, reflected in significant improvements of several 2D and 3D indices. Large increments in bone volume fraction are one of the most striking and consistent findings in ovariectomized rats treated with PTH and are associated with increased biomechanical strength.(36,37) Previous reports on changes in cancellous bone volume in humans after PTH treatment have, however, been conflicting. Three previous studies of human iliac crest biopsy specimens using traditional 2D bone histomorphometry reported an increase,(13,14,17) but two other studies were unable to show an increase in cancellous bone volume.(15,16) These discrepancies are probably caused by the limited number of samples available for assessment in these studies. Treatment with teriparatide for 6 months was found to induce a 2-fold increase in bone volume fraction,(13) and a 50% increase was observed in estrogenized women after 1 year of teriparatide treatment.(14) Also, two indices reflecting connectivity of the trabecular network, marrow star volume assessed by 2D methods and connectivity index assessed by 3D μCT, showed significant improvement.

Increased 2D trabecular connectivity after teriparatide treatment has been demonstrated in rats(38) and monkeys.(39) Greater 2D trabecular connectivity has also been found in patients with mild primary hyperparathyroidism when they were compared with age- and sex-matched controls.(40) The increased 3D connectivity density after teriparatide treatment in these studies support data in a recent report where a similar observation was made based on a limited number of iliac specimens.(16) Improvement of connectivity indices does not directly prove reconnection of perforated trabeculae. Trabecular thickening may also cause a decrease of marrow star volume, and the connectivity index may actually increase in the early phases of perforation of trabecular plates in young bone.(41) Another possibility of increasing connectivity density is intratrabecular tunneling(39,42) or trabecular perforation.(43) Intratrabecular tunneling in iliac crest and vertebral sections has been described in monkeys after teriparatide treatment,(39) possibly as a remodeling mechanism to maintain trabecular thickness. Tunneling of thickened individual trabeculae would convert them into multiple trabeculae, resulting in a normalization of trabecular thickness but an increase in the number of 3D spatial connections. However, when analyzing our images obtained from the μCT analysis, increased connectivity and a reversal of rod-like structure to a plate-like pattern can be observed. This is also reflected quantitatively in the decrease in structure model index in biopsy specimens from women treated with teriparatide.

The structural changes accompanying age-dependent bone loss and osteoporosis are loss of trabecular connectivity caused by trabecular perforations and cortical thinning.(44–46) The trabecular perforations also cause a transition of trabecular morphology from the plate-like appearance in younger individuals to a more rod like appearance. Our qualitative and quantitative data show that teriparatide not only increases bone volume but also reverses the osteoporotic changes in trabecular morphology and cortical geometry. Earlier studies on human vertebral specimens have demonstrated that the combination of trabecular bone volume fraction with other 2D structural parameters improves the prediction of bone strength in a multiple regression model.(47) At this time, only limited data are available for humans, although the association of bone volume fraction with bone biomechanical properties is well documented.(19)

One study in ovariectomized rats on estrogen replacement therapy reported that the SMI and 3D connectivity density predicted vertebral compressive strength. Another aging study in pigs has demonstrated that inclusion of connectivity density with bone volume in a regression analysis improved the prediction of both maximum vertebral stress from compressive testing and apparent modulus from finite element modeling.(48)

In our analysis, the changes of more simple 2D indices pertaining to cancellous bone structure, Tb.N, Tb.Th, and Tb.Sp, did not reach significance after teriparatide treatment. However, more stereologically correct indices, like marrow star volume and μCT-based 3D indices revealed significant changes, further corroborating the superiority of these techniques for structural analysis of small samples such as bone biopsy specimens.

When material properties change, such as in rats on low-dose long-term sodium fluoride treatment, structural parameters fail to predict biomechanical properties.(49) However, for teriparatide, clear correlations between improvement of structure and biomechanical properties have been reported in monkeys.(30,50) Using a combination of QCT, microfinite element analysis, and compression testing, significant improvements in biomechanical competence of vertebrae from monkeys treated with teriparatide for 6 months have been demonstrated. Thus, teriparatide does not negatively impact the material properties of bone that would offset the positive effects on bone architecture.

In conclusion, teriparatide treatment stimulates both trabecular and cortical bone formation, resulting in increased cortical thickness and cancellous bone volume, improved cancellous bone connectivity, and a shift from rod-like to plate-like trabecular morphology. These changes constitute a reversal of osteoporotic bone structural changes and explain the pronounced decrease in vertebral and nonvertebral fracture rates observed after teriparatide treatment.

Acknowledgements

The authors thank Karen V Pinette for manuscript preparation and Anette Baatrup for preparation and sectioning of bone samples. This study was supported by Eli Lilly and Company, Indianapolis, Indiana, USA.

Ancillary