Teriparatide Increases Bone Formation in Modeling and Remodeling Osteons and Enhances IGF-II Immunoreactivity in Postmenopausal Women With Osteoporosis

Teriparatide Increases Bone Formation in Modeling and RemodelingOsteons and Enhances IGF-II Immunoreactivity in Postmenopausal

Women With Osteoporosis*

Yanfei L Ma,1 Qingqiang Zeng,1 David W Donley,1 Louis-Georges Ste-Marie,2 J Christopher Gallagher,3

Gail P Dalsky,1 Robert Marcus,1 and Erik Fink Eriksen1

ABSTRACT: Transiliac bone biopsies were obtained from 55 women treated with teriparatide or placebo for12–24 months. We report direct evidence that modeling bone formation at quiescent surfaces was present onlyin teriparatide-treated patients and bone formation at remodeling sites was higher with teriparatide thanplacebo.

Introduction: Recombinant teriparatide [human PTH(1-34)], a bone formation agent for the treatment ofosteoporosis when given once daily subcutaneously, increases biochemical markers of bone turnover andactivation frequency in histomorphometry studies.Materials and Methods: We studied the mechanisms underlying this bone-forming action of teriparatide at thebasic multicellular unit by the appearance of cement lines, a method used to directly classify surfaces asmodeling or remodeling osteons, and by the immunolocalization of IGF-I and IGF-II. Transiliac bone biopsieswere obtained from 55 postmenopausal women treated with teriparatide 20 or 40 �g or placebo for 12–24months (median, 19.8 months) in the Fracture Prevention Trial.Results: A dose-dependent relationship was observed in modeling and mixed remodeling/modeling trabecularhemiosteons. Trabecular and endosteal hemiosteon mean wall thicknesses were significantly higher in bothteriparatide groups than in placebo. There was a dose-dependent relationship in IGF-II immunoreactivestaining at all bone envelopes studied. The greater local IGF-II presence after treatment with teriparatide mayplay a key role in stimulating bone formation.Conclusions: Direct evidence is presented that 12–24 months of teriparatide treatment induced modeling boneformation at quiescent surfaces and resulted in greater bone formation at remodeling sites, relative to placebo.J Bone Miner Res 2006;21:855–864. Published online on May 8, 2006; doi: 10.1359/JBMR.060314

Key words: teriparatide, transiliac bone biopsy, bone formation, modeling, osteoporosis

INTRODUCTION

OSTEOPOROTIC FRACTURES ARE a biomechanical conse-quence of a decrease in bone quality; that is, the com-

ponents of bone strength, such as structural integrity ormicroarchitectural deterioration that occurs with increasingage that are not measured by bone densitometry.(1–3)

Therapeutic drugs for the treatment of osteoporosis in-crease bone strength, as shown by a decrease in the inci-dence of fragility fractures, through their effects on boneturnover.(4) Biochemical markers of bone formation havemore rapid and larger increases than resorption markerswithin the first 3 months of teriparatide therapy, suggestingan early imbalance of bone turnover in favor of formationthat remain positive over the first 6–12 months of treat-ment.(5–7) Markers of bone turnover may reflect underlyingchanges in bone histology and provide indirect evidence ofa dissociation of bone formation and resorption.(8–10)

The predominant mechanism of bone turnover in adultsis bone remodeling, the cellular basis for age-related boneloss when the processes of bone resorption and formationbecome unbalanced. Frost hypothesized that modeling mayoccur throughout life, although modeling activity is less fre-quent after skeletal maturity.(11,12) Antiresorptive agentsreduce bone turnover, decreasing the size of the remodelingspace and the number of basic multicellular units (BMUs),resulting in a preservation of bone microarchitecture.(13,14)

*This paper was previously presented at the 1st Joint Meeting ofthe International Bone and Mineral Society/Japanese Society ofBone and Mineral Research, June 3–7, 2003, Osaka, Japan, the 9thBath Conference on Osteoporosis, National Osteoporosis Society,June 23–26, 2003, Bath, United Kingdom, the 30th European Sym-posium on Calcified Tissues, May 8–12, 2003, Rome, Italy, andAdvances in Skeletal Anabolic Agents and the Treatment of Os-teoporosis, May 24–25, 2004, Bethesda, MD, USA.

Drs Ste-Marie and Gallagher have received consultant fees fromEli Lilly and Company. Drs Ma, Zeng, Donley, Dalsky, and Mar-cus are employees of Eli Lilly and Company; Dr Eriksen was anemployee of Eli Lilly and Company at the time the work was done.Dr Eriksen is now an employee of Novartis.

1Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, Indiana, USA; 2CHUM Hopital Saint-Luc, Montreal, Quebec,Canada; 3Bone Metabolism Unit, Creighton University, Omaha, Nebraska, USA.

JOURNAL OF BONE AND MINERAL RESEARCHVolume 21, Number 6, 2006Published online on May 8, 2006; doi: 10.1359/JBMR.060314© 2006 American Society for Bone and Mineral Research

855

JO506359 855 864 June

Recombinant teriparatide {human PTH(1-34) [hPTH (1-34)]}, currently the only bone-forming osteoporosis drugavailable for clinical use,(15,16) increases bone turnover witha greater stimulation of formation than resorption. Teripa-ratide increases the size of the remodeling space and thenumber of BMUs, resulting in a positive bone balance andimproved macro- and microarchitecture.(10,17–19) Changesin bone microarchitecture with teriparatide treatment, as-sociated with an increase in bone strength, include in-creased trabecular bone volume, bone width, connectivitydensity, improved morphology with a shift toward moreplate-like structure, and increased cortical thick-ness.(17,18,20–23)

Mechanisms for improvements in microarchitecture bybone-forming agents may be through modeling or remod-eling.(24) In short-duration treatment (1–6 months) studieswith bone-forming agents in animals(24–26) and hu-mans,(20,22,23,27) bone modeling plays an important role.Reeve et al.(20) suggested that 6 months treatment withteriparatide caused dissociation between formation and re-sorption resulting in a positive bone balance in patients withosteoporosis. Hodsman and Steer(22) hypothesized that newbone formation occurred on previously quiescent surfacesafter 28 days of teriparatide treatment in women with os-teoporosis. This hypothesis was supported by indirect evi-dence from increases in activation frequency and doubletetracycline-labeled surfaces and is consistent with the well-known effects of teriparatide in increasing markers of boneformation.(15,28)

The appearance of cement lines and collagen fiber ori-entation are used to directly classify surfaces as modeling(i.e., new bone formation occurring on a quiescent surfacewithout preceding osteoclast bone resorption).(29) At mod-eling sites, cement lines are smooth, with parallel collagenfiber orientation, absent of prior osteoclastic activity.(29) Inthe remodeling process, the cement line is scalloped, indica-tive of the osteoclastic activity that precedes bone forma-tion in a BMU.(29)

Dempster et al.(23) reported that 28 days of teriparatidetreatment directly stimulated bone formation without priorresorption on both trabecular and endocortical bone sur-faces in postmenopausal women with osteoporosis, basedon the presence of scalloped or smooth cement lines. Theyfound modeling-based formation on trabecular and endo-cortical surfaces in teriparatide-treated patients, but not inbiopsy samples from controls.(23) Using the technique ofquadruple labeling to obtain longitudinal results from asingle biopsy after 7 weeks of teriparatide treatment, Lind-say et al.,(27) in their abstract, suggested that bone forma-tion occurred either on previously quiescent surfaces or byosteoblast spillover from remodeling sites onto previouslyquiescent surfaces.

IGFs play an important role in the regulation of skeletalremodeling by upregulating osteoblast and osteoclast activi-ties associated with acquisition of BMD.(30–32) Growth fac-tors may also mediate the effects of PTH in stimulating newbone formation.(33) Although the age-related decline inBMD and bone strength is well known, less recognized as arisk factor for osteoporotic fracture is the decline in serumIGF concentrations with age.(34,35)

The aim of this study was to investigate the workingmechanism of the bone-forming action of teriparatide byanalyzing the occurrence and dimensions of modeling andremodeling osteons and IGF-I or IGF-II expression intransiliac bone biopsies obtained from patients after either12 months of treatment or at the end of treatment (range,19–24 months) in the Fracture Prevention Trial.(15)

MATERIALS AND METHODS

Subjects

Transiliac bone biopsies were obtained at study baselinefrom 102 women who volunteered for the biopsy substudyat selected study sites. The volunteers were randomly as-signed to provide a second biopsy from the contralateralside after either 12 months of treatment or at the end of thestudy (median, 22.0 months; range, 19–24 months). Ap-proximately one third of the women were randomized tohave the second biopsy after 12 months of treatment andtwo thirds after 24 months of treatment. More patients wereassigned to the longer-duration time to improve the likeli-hood of acquiring the desired number of paired biopsiesadequate for evaluation. The volunteers were a subset ofthe 1637 postmenopausal women with osteoporosis en-rolled in the randomized, multicenter, double-blind, pla-cebo-controlled Fracture Prevention Trial.(15) To enroll inthe Fracture Prevention Study, a patient’s laboratory valuefor 25-hydroxyvitamin D had to be between the lower limitof normal and three times the upper limit of normal atbaseline. Normal was defined by the central laboratory ref-erence range. All of the patients in this substudy had 25-hydroxy and 1,25 dihydroxyvitamin D levels above thelower limit of normal at baseline.

Participants were treated with once-daily self-injectionsof either placebo or teriparatide 20 or 40 �g. Details of thisstudy have been published.(15) Baseline BMD measure-ments of the lumbar spine and femoral neck were obtainedby DXA as previously described.(15) Biochemical markersof bone formation (serum bone-specific alkaline phospha-tase [BALP], serum procollagen type I C-terminal propep-tide [PICP]) and resorption (urinary N-terminal telopep-tide cross-linked collagen type I [NTX]) were measured atbaseline on fasting samples.

No evidence of modeling was found in the baseline sec-tions obtained before treatment; therefore, the analysis andcomparisons were restricted to samples obtained aftertreatment. Of the 61 follow-up biopsies obtained, 52samples were of sufficient quality for histomorphometricalanalyses for modeling/remodeling activity. Nine biopsyspecimens were excluded from histomorphometry becausethey were crushed or broken (six of nine) or did not havepair-matched sections for both cement line and polarizedlight quantification (three of nine). Immunohistochemistrywas performed on a total of 39 samples from among 55unbroken biopsy specimens; 13 samples randomly selectedfrom the intact specimens for each experimental group (pla-cebo and teriparatide 20 and 40 groups). Thirty-six biopsieswere used for both histomorphometric and immunohisto-chemistry analyses. The technician who performed theanalysis was blinded to the treatment group assignment.

MA ET AL.856

Biopsies used for the histomorphometry modeling andremodeling analyses were distributed as follows---placebo:12 months (n � 7), study endpoint (n � 13); teriparatide 20�g: 12 month (n � 6), endpoint (n � 13); teriparatide 40�g: 12 month (n � 5), endpoint (n � 8). Immunocyto-chemistry analyses were performed on samples obtained at12 months (placebo, n � 5; teriparatide, 20 �g, n � 4;teriparatide 40 �g, n � 6) and study endpoint (placebo, n� 8; teriparatide 20 �g, n � 9; teriparatide 40 �g, n � 7).

The Institutional Review Board for Research InvolvingHuman Subjects at each participating center approved theprotocols for the Fracture Prevention Trial and the biopsysubstudy. Volunteers gave written informed consent beforeparticipation in the treatment and biopsy studies.

Treatment and biopsy schedule

Participants self-administered a once-daily injection ofplacebo for 2 weeks and were randomly assigned to receiveplacebo or 20 �g or 40 �g of teriparatide daily for 3 years.The Fracture Prevention Trial(15) was terminated early be-cause of a finding of osteosarcoma in rats(36,37); therefore,the treatment duration was shorter than originally planned.Participants received daily supplements of 1000 mg of el-emental calcium and 400–1200 IU of vitamin D3 for theduration of the trial.

The transiliac bone biopsies were carried out using amodified Bordier needle or similar large bore (6–8 mm)trephine system after in vivo double tetracycline labeling(tetracycline HCL 250 mg, four times a day) given orally ina 3:12:3-day sequence, as described by Jiang et al.(17) Themethods for biopsy specimen evaluations have been pub-lished previously.(38,39) All measured and derived variableswere expressed according to standard nomenclature recom-mended by the American Society of Bone and Mineral Re-search nomenclature committee.(39)

Histomorphometry

Biopsies were dehydrated in graded ethanol and embed-ded in methylmethacrylate at −20°C. From each block, sev-eral 8-�m-thick undecalcified sections were cut with 100-�m spacing and prepared for different staining methods orleft unstained. Of each section pair, the unstained sectionwas used for tetracycline fluorochrome and polarized lightanalyses. The toluidine blue–stained sections were used toview cement lines, osteoblasts, and wall thickness, whereasGoldner’s trichrome–stained sections were used to measuretissue area, bone area, and bone surface under transmittedlight. All biopsy measurements were made by laboratorypersonnel blinded to individual treatment assignments.

Active bone-forming, tetracycline-labeled osteons on tra-becular and endocortical surfaces were studied using polar-ized light for collagen orientation and transmitted light forcement line stains (Fig. 1). Modeling and remodeling analy-ses were performed on the trabecular structure hemios-teons of the entire trabecular tissue area. Only those tra-becular structural units containing either tetracycline labelor osteoblast surfaces as evidence for active bone formationat the time of biopsy were analyzed. The osteons were clas-sified according to the presence of smooth (modeling) or

scalloped (remodeling) cement lines. A remodeling site wasdefined by the presence of a scalloped cement line withinterrupted collagen fibers indicating that formation fol-lowed previous bone resorption or osteoclastic activity.(29)

In bone-forming modeling sites, the cement lines aresmooth, showing evidence that no osteoclastic bone resorp-tion preceded bone formation. The modeling unit was fur-ther assessed under polarized light to ensure that collagenfibers in the whole unit followed the same orientation asthose of the adjacent bone tissue. Bone-forming units withthe presence of very short scalloped (less than one sixth ofthe total cement line length) and long smooth cement lineswere classified as prolonged remodeling sites or mixed re-modeling-modeling hemiosteons (Fig. 1). To avoid the cut-ting orientation effect, all the modeling and mixed remod-eling-modeling units were identified in longitudinal cuttingorientation and further confirmed by polarized light used toexamine the collagen fiber patterns in the whole unit. Thepercentage of each type of the hemiosteons from the totalnumbers of formation osteons was calculated.

Bone area, bone surface, tissue area, and trabecular andendocortical mean wall thickness were analyzed semiauto-matically by a direct tracing method using a Digitizing Im-age Analysis System (DIAS) consisting of an epifluorescentmicroscope and a digitizing pad (Summagraphic, Fairfield,CT, USA) coupled to an Apple Macintosh SE computer,using histomorphometry software (KSS Scientific Consult-ants, Magna, UT, USA).(40) Wall thickness (�m) was mea-sured on toluidine blue–stained sections as the average dis-tance between cement lines and quiescent trabecularsurfaces, corrected for obliquity. Wall thickness was mea-sured only on tetracycline-labeled osteons covered with lin-ing cells, indicating finished (quiescent) bone remodelingunits formed during the treatment period. All measure-ments were performed at ×100 magnification on the entiretrabecular and endocortical envelopes. The parameterswere normalized to bone, surface, and tissue reference. Onaverage, 16–30 formation osteons were measured per sec-tion (range, 6–68). Six to 16 such finished remodeling os-teons were used for identification of wall thickness in eachbiopsy sample.

Three types of hemiosteons were observed (Fig. 1): re-modeling hemiosteons (Figs. 1A and 1B) identified by scal-loped reversal lines that interrupt surrounding collagen fi-bers, indicating prior resorption; mixed remodeling-modeling hemiosteons (Figs. 1C and 1D) identified by amixture of very short scalloped cement lines and longsmooth cement lines; and modeling hemiosteons (Figs. 1Eand 1F) identified by smooth cement lines with parallelcollagen fibers without interruption of adjacent bone tissue.

Immunocytochemistry

To maximize immunoreactivity, bone biopsies were em-bedded in methylmethacrylate at –20°C.(41) Unstained7-�m sections were deplasticized in two successive 20-minute washes of xylene, followed by two washes in ethanoland a brief wash in PBS containing 0.1% Tween 20 (PBS-T), pH 7.4. Sections were treated with 1% acetic acid for 10minutes to expose antigenic sites. Before the addition of the

TERIPARATIDE-INDUCED MODELING BONE FORMATION 857

primary antibody, nonspecific tissue binding was blockedby incubating the tissue section in 10% BSA for 30 minutesat room temperature. The collagen antibodies used in thisstudy were supplied by Dr Juha Risteli, from the University

of Oulu, Finland, and tested by him to have <1% cross-reactivity with other collagen species. Subsequently, sec-tions were incubated overnight with primary antibodyagainst human IGF-I, II, bone morphometric protein

FIG. 1. Remodeling hemiosteons. (A) Unstained fluorescence light image and (B) transmitted light for toluidine blue staining imagespecific for cement lines. Newly formed lamellar bone is indicated by the red arrows. The presence of double label is evident here.Reversal or scalloped cement line in the adjacent bone or interrupted collagen fiber orientation from adjacent bone tissue underlyingthe forming site outlined by black arrows, indicating previous bone resorption surface, evidenced by scalloped curved fine lines(62-year-old woman treated with teriparatide 40 �g). Mixed remodeling--modeling hemiosteons. (C) Unstained fluorescence lightimage and (D) transmitted light for toluidine blue staining image specific for cement lines. Newly formed lamellar bone is indicated bythe red arrows. Very short scalloped cement line and a long smooth cement line in the adjacent bone underlying the forming siteoutlined by black arrows, indicating previous bone resorption surface, evidenced by scalloped curved fine lines. Arrows indicate thatformation was initiated on the resorption site and then extended beyond the previous resorption surface. Smooth cement lines areindicated by the yellow arrows. The inset in C shows the presence of lamellar bone (73-year-old woman treated with teriparatide 40�g). Modeling hemiosteons. (E) Unstained fluorescence light image and (F) polarized light. Newly formed lamellar bone is indicatedby the red arrows. Smooth cement lines and parallel collagen fiber orientation as the adjacent bone tissue, without interruption,indicating absence of previous bone resorption (yellow arrows; 73-year-old woman treated with 40 �g teriparatide).

MA ET AL.858

Fig 1 live 4/C

(BMP)-2, TGF�1, and type I collagen. The IGF-I andIGF-II monoclonal antibodies were obtained fromBIOSOURCE (Camarillo, CA, USA); TGF�1, rabbit poly-clonal, and BMP-2, goat, polyclonal, affinity-purified anti-bodies were from Santa Cruz Biotechnology (Santa Cruz,CA, USA). After rinsing, the sections were incubated withperoxidase labeled secondary biotin-avidin antibodies for 1h and subsequently developed in ethylcarbazole. Afterwashing in buffer, the chromagen, diaminobenzidine wasapplied for 5 minutes followed by a counterstain with May-er’s hematoxylin.

Negative controls included substituting the primary anti-sera with preimmune sera from the same species, and omit-ting the primary antibody. All controls revealed the ex-pected negative results. The total length of IGF-II–positivestained cement lines was first measured in trabecular andcortical bones using the same image system describedabove. The length was normalized to bone area and surfaceperimeter for comparison. The entire trabecular bone in thesample for IGF staining was quantified.

Statistical analysis

Demographics and relevant biopsy and markers of boneturnover variables at baseline were summarized by groupmean and SD. To compare the baseline characteristicsamong the three treatment groups, each pairwise compari-son was performed. Two-sample t-tests were used for age,years postmenopausal, body mass index, lumbar spine, andfemoral neck T score. The Wilcoxon rank sum test was usedfor the bone markers: NTX, BALP, and PICP. SeparateFisher’s exact tests were performed for the number of pa-tients with zero, one, and at least two prevalent fractures.

The postbaseline data were similar between the twotime-points (12 months and study endpoint) for each of thetreatment groups; therefore, the postbaseline biopsies werepooled for analyses. For each parameter, separate two-sample tests were performed to compare teriparatide 20 �gversus placebo and to compare teriparatide 40 �g versusplacebo. The Wilcoxon rank sum test was used for eachcomparison; p values for each comparison were obtained

using Monte Carlo simulation because of the small samplesize. � level of p < 0.05 was the criterion for statisticalsignificance with no adjustment for multiplicity. All analy-ses were performed using SAS version 8 statistical software(SAS Institute, Cary, NC, USA). Data are presented asmean ± SD in the text and tables and as mean ± SE in thefigures.

RESULTS

Demographics

There were no statistically significant differences in base-line characteristics among groups (Table 1). Baseline de-mographics for the biopsy substudy did not differ fromthose of the Fracture Prevention Trial population.(15)

HistomorphometryThere was no woven bone, osteomalacia, or other histo-

logical abnormality observed in the postbaseline biopsiesfor any of the patients in this substudy. Three types ofhemiosteons were observed in biopsy samples after teripa-ratide treatment: remodeling hemiosteons (Figs. 1A and1B); mixed remodeling-modeling hemiosteons (Figs. 1Cand 1D); and modeling hemiosteons (Figs. 1E and 1F).

No modeling hemiosteons or mixed remodeling-modeling hemiosteons were found on trabecular surfaces inplacebo-treated patients. Compared with placebo, a signifi-cantly greater percentage of active, tetracycline labeledmodeling hemiosteons on trabecular surfaces was seen inwomen treated with teriparatide 40 �g (3.8%, p < 0.001),but not in women treated with 20 �g (0.4%, Fig. 2). Therewere also significantly greater percentages of mixed remod-eling/modeling osteons on trabecular surfaces for bothteriparatide groups (teriparatide 20 �g, 2.4%; teriparatide40 �g, 3.9%, p < 0.05 versus placebo for both teriparatidegroups; Fig. 2). There were 6–68 such formation osteons ineach biopsy samples: 0–6 modeling; 6–65 remodeling; 0–5mixed remodeling-modeling.

Mean wall thickness of trabecular hemiosteons was sig-nificantly greater than placebo (p < 0.05) in teriparatide 20and 40 �g (Fig. 3). The difference was more pronounced on

TABLE 1. BASELINE CHARACTERISTICS OF PATIENTS*

Placebo (n = 22) TPTD20 (n = 19) TPTD40 (n = 14)

Age (years) 67.6 ± 6.3 68.2 ± 5.6 67.7 ± 7.6Years postmenopausal 20.2 ± 8.0 21.4 ± 8.5 22.1 ± 9.9Body mass index (kg/m2) 27.1 ± 4.7 26.7 ± 3.8 26.0 ± 4.4Lumbar spine T score −2.2 ± 1.5 −2.9 ± 1.2 −2.4 ± 1.5Femoral neck T score −2.1 ± 0.7 −2.3 ± 0.7 −2.2 ± 0.7NTx/Cr (nmol BCE/mmol) 46.0 ± 23.5 49.5 ± 23.6 41.7 ± 20.2BAP (�g/liter) 12.8 ± 6.3 16.7 ± 9.9 12.1 ± 5.0PICP (�g/liter) 116.7 ± 34.1 105.3 ± 26.9 133.8 ± 83.2No. of prevalent vertebral fractures [N (%)]

No fractures 3 (13.6%) 1 (5.3%) 01 fracture 6 (27.3%) 6 (31.6%) 4 (28.6%)�2 fractures 13 (59.1%) 12 (63.2%) 10 (71.4%)

Values are mean ± SD or as N (%)* No significant differences between teriparatide doses and placebo.TPTD20, TPTD40, teriparatide 20 or 40 �g; NTx/Cr, creatinine corrected urinary N-terminal telopeptide cross-linked collagen type I; BAP, serum bone

alkaline phosphatase; PICP, serum procollagen type I C-terminal propeptide.

TERIPARATIDE-INDUCED MODELING BONE FORMATION 859

the endocortical surfaces in both teriparatide groups com-pared with placebo (Fig. 3).

Immunocytochemistry

Immunoreactivity for IGF-I, IGF-II, BMP-2, TGF�1,and collagen I was detectable in bone matrix with prefer-ential localization in cement lines, reversal lines of bone,and the outer lining cell layer of bone remodeling compart-ments. Among them, collagen I showed a high signal andqualitative analysis, suggesting that teriparatide treatmentmarkedly upregulated the expression of collagen I on bonematrix, but the staining was too poorly delineated to enablequantification (Fig. 4). Only IGF-II showed sufficientlyhigh levels and had clear borders to warrant quantification.IGF-II staining was evaluated in periosteal, cortical, endo-cortical, and trabecular bone (Table 2). The highest expres-sion of IGF-II was found in cement lines (Fig. 5). Whenenumerating the length of cement lines with positive IGF-IIimmunoreactivity, we found that both teriparatide treat-ment groups upregulated IGF-II expression compared withplacebo treatment (Table 2). The in situ expression of IGF-II, per unit cortical area, cortical osteon number, trabecularsurface, and trabecular area, was dose-dependently greaterwith teriparatide treatment than with placebo treatment(Table 2). IGF-II expression was also significantly greater

on periosteal surfaces for both teriparatide groups com-pared with placebo (Table 2).

DISCUSSION

Our findings of modeling and mixed remodeling-modeling hemiosteons in postmenopausal women with os-teoporosis provide direct evidence that 12–24 months ofrecombinant teriparatide treatment induced new bone for-mation at quiescent surfaces. That findings of modelingwere evident after 12–24 months of teriparatide treatmenthighlights a distinction of this research from the work ofHodsman and Steer,(22) Dempster et al,(23) and Lindsay etal.,(27) who reported that an early effect determined afterjust a month of teriparatide treatment was increased bonemodeling. Hodsman and Steer(22) hypothesized that the in-creases in double tetracycline-labeled surfaces were com-patible with new bone formation occurring on previouslyquiescent surfaces after 28 days of teriparatide therapy inpostmenopausal women. Dempster et al.(23) reported thatmodeling-based formation activity on trabecular surfaceswas found after 28 days of teriparatide treatment. Lindsayet al.(27)observed that 70% of new bone formation was re-modeling based, and 30% was modeling based. The greaterproportion of remodeling-based formation is likely a func-tion of the longer duration of teriparatide treatment in ourstudy.

FIG. 4. Effect of teriparatide treatment on in situ staining of type I collagen in trabeculae detected by immunohistochemistry.Transiliac bone biopsy sections were counterstained with Mayer’s hematoxylin after incubated with primary antibody against humancollagen I. Greater intensity of staining in teriparatide-treated women than in placebo is apparent. (Women treated with placebo [62years old] or teriparatide 20 �g [66 years old] or 40 �g [58 years old], respectively.)

FIG. 2. Percent of modeling, mixed remodeling-modeling, andremodeling hemiosteons in trabeculae during teriparatide treat-ment (mean ± SE). No modeling or mixed remodeling-modelinghemiosteons were observed in the placebo group. Bars to the rightof the dotted line (remodeling) use the right axis. ap < 0.001 vs.placebo.

FIG. 3. Mean wall thickness in cancellous and endocorticalhemiosteons (mean ± SE). ap < 0.05 vs. placebo.

MA ET AL.860

Fig 4 live 4/C

Bone turnover can be measured by activation frequency,which is the probability that a new cycle of remodeling willbe initiated,(39) that is, an index to show how many newBMUs (remodeling packet) are initiated in a given time.Activation frequency is increased during the early monthsof treatment, but by 18 months, activation frequency interiparatide-treated patients was comparable with that ofuntreated patients.(10) With continued treatment, althoughactivation frequency may have returned to untreated levels,a positive bone balance is maintained, as indicated byhigher mean wall thickness.(42) BMD continues to increasethroughout the teriparatide treatment,(15) and we suggestthat this gain in BMD occurs through a combination ofmodeling and remodeling.

Biochemical markers of bone turnover provide support-ing evidence that, during the first month of teriparatidetherapy, new bone formation may be modeling-based be-cause only formation markers show responses during thisearly phase.(9,15,28,43) After 1 month, resorption markerssteadily increase, although they remain proportionatelylower than formation markers, suggesting that bone forma-tion gradually becomes more remodeling-based.

The classification of cement lines as smooth or scallopedwas used to determine whether modeling- or remodeling-based new bone formation occurred, as described in otherstudies.(12,27,29) As discussed recently by Lindsay et al,(27)

although smooth cement lines are associated with model-ing-based new bone formation, this assumption is unlikelyto be confirmed by research findings. Lindsay et al. postu-lated that a lack of scalloped cement lines may result fromthe resorption of lamellar bone by osteoclasts. Another ex-planation is that smooth cement lines occur when osteo-clasts resorb along collagen fibers. Nevertheless, we andothers found that new bone formation osteons with smoothcement lines are uncommon under normal circumstances.

We also report the presence of new structures in recom-binant teriparatide-treated bone, mixed remodeling-modeling osteons, consisting of very short scalloped andlong smooth cement lines. We did not observe such a mod-eling-dominant structure in placebo-treated participants.We interpret this structure as an indication that formation isinitiated at active remodeling sites on the resorption surfaceand then extended beyond the previous resorption surfaceduring the bone formation period with teriparatide treat-

TABLE 2. EFFECT OF TERIPARATIDE TREATMENT ON IN SITU EXPRESSION OF IGF-II IN TRABECULAR AND CORTICAL BONE AND ON

THE PERIOSTEAL AND ENDOCORTICAL SURFACES

Placebo TPTD20* TPTD40*

IGF-II perimeter/trabecular area (mm/mm2) 1.3 ± 0.8 2.4 ± 1.4NS

2.8 ± 2.0p�0.029

IGF-II perimeter/cortical area (mm/mm2) 0.4 ± 0.3 1.1 ± 0.6p < 0.01

1.5 ± 1.1p < 0.01

IGF-II perimeter/trabecular surface (%) 7.2 ± 4.5 15.7 ± 10.8p�0.044

18.8 ± 11.8p < 0.01

IGF-II perimeter/periosteal surface (%) 13.6 ± 7.3 28.7 ± 10.8p < 0.01

28.9 ± 11.8p < 0.01

IGF-II perimeter/osteon (mm/#) 0.08 ± 0.06 0.20 ± 0.09p < 0.01

0.23 ± 0.16p < 0.01

IGF-II perimeter/endocortical surface 11.1 ± 5.4 25.1 ± 15.7p < 0.01

18.5 ± 8.5p�0.015

Values are mean ± SD.p values from two-sample Wilcoxon rank sum test were obtained using Monte Carlo simulation.*Teriparatide 20 or 40 �g versus placebo.TPTD20, TPTD40, teriparatide 20 or 40 �g; NS, not statistically significant, p > 0.05.

FIG. 5. Microimages showing the effect ofteriparatide treatment on immunocytochem-ical staining of IGF-II in trabeculae and peri-osteum. Transiliac bone biopsy sections werecounterstained with Mayer’s hematoxylin af-ter incubated with primary antibody againsthuman IGF-II. Teriparatide treatment re-sulted in a dose-dependent increase in immu-nocytochemical staining of IGF-II on bonematrix. Black arrows indicate staining on ma-jor cement lines. (Women treated with pla-cebo [69 years old] or teriparatide 20 �g [70years old] or 40 �g [73 years old], respec-tively.)

TERIPARATIDE-INDUCED MODELING BONE FORMATION 861

Fig 5 live 4/C

ment. The appearance of this structure with modeling areassurrounding the resorption lacunae suggests that teripa-ratide causes extension or “overfilling” of trabecular hemi-osteons beyond the boundaries of the original remodelingsite onto a quiescent surface, thus greatly enhancing thepositive remodeling balance at each remodeling unit.(18,44)

In agreement with Dempster et al.(23) and Lindsay etal.,(27) who found an absence of modeling in their controlpatients, we likewise found no typical modeling hemios-teons in our placebo-treated patients. In contrast, Kobaya-shi et al.(12) reported modeling activity in the iliac crestsamples collected during the total hip arthroplasty, al-though the extent of minimodeling was very low (1% ofbone volume and 2% of bone surface). The finding of sig-nificantly more modeling and mixed remodeling-modelingosteons in teriparatide-treated than in placebo-treated pa-tients supports the hypothesis that teriparatide increasesand induces bone formation at Frost’s intermediary orga-nizational level (modeling and remodeling).(45)

Wall thickness is a measure of the final bone balanceafter a remodeling cycle has finished. Increased mean wallthickness, indicative of a positive bone balance,(46) washigher in the teriparatide-treated groups than in placebo, asshown previously.(18,27,42,44) We confirmed greater meanwall thickness of endosteal osteons relative to placebo aspreviously reported,(23,42) and also observed a greater meanwall thickness of cancellous hemiosteons relative to placebothat may indicate stimulation of osteoblastic synthesis ofnew bone matrix, especially type I collagen. The apparentincrease in collagen I levels associated with teriparatidetreatment in our study supports this hypothesis.

Although IGF-I has been primarily established in ro-dents as the dominant mediator, our results suggest thatIGF-II is more abundant than IGF-I in human bone, inagreement with Mohan and Baylink,(47) who reported thatthe growth factor showing the highest concentrations in hu-man bone matrix is IGF-II. Recently, Pepene et al.(48) re-ported that some antiresorptive treatments (estrogentherapy and calcitonin) had no direct influence on serum orbone matrix growth factor levels. Boonen et al.(35) foundthat IGF-II concentrations were 40% lower in elderlywomen with hip fractures compared with healthy, age-matched controls. Although periosteal bone formation oc-curs at a negligible level during adulthood, particularly inwomen, our results suggest that teriparatide may reactivateperiosteal bone formation. From a biomechanical view, anincrease in periosteal apposition contributes to improvedbone strength and improved resistance to fracture.(49,50)

IGF-II is a known stimulator of osteoblastic activity andhas been positively associated with osteoblast surface.(51,52)

Our study showed for the first time that IGF-II immuno-staining was twice as high on periosteal surfaces in teripa-ratide-treated patients than in placebo-treated patients,which may play a key role in the creation of a positive bonebalance and improved trabecular and cortical bone archi-tecture previously shown in these biopsies. Our inability toshow significant changes in other growth factors such asBMP-2, and TGF�1 may signify either an earlier responseor lesser involvement by these factors in the bone-formingresponse.

There is a limitation in characterizing bone structureunits by microscopy in histological sections. Although wecall the hemiosteons with smooth cement lines “modeling”units, they might conceivably be part of a remodeling site,where we have sectioned through the more peripheralparts. Thus, in this study, all the modeling and mixed re-modeling-modeling units were identified in longitudinalcutting orientation and further confirmed by polarized lightto examine the collagen fiber patterns in the whole unit. 3Dconstruction of these hemiosteons would help to confirmwhether they are true modeling or remodeling-modelingunits.

In summary, these results provide the first direct evi-dence that teriparatide treatment of 12- to 24-month dura-tion induces modeling bone formation at quiescent surfacesand that bone formation at remodeling sites in teriparatide-treated groups is greater than in the placebo group. Thesemechanisms may contribute to the improvement of trabec-ular and cortical architecture shown after teriparatide treat-ment.

ACKNOWLEDGMENTS

This study was funded by Eli Lilly and Company. Theauthors thank Juha Risteli, MD, PhD, at the University ofOulu, Finland, who supplied the collagen antibodies andchecked their specificity, Webster Jee, PhD, at the Univer-sity of Utah for critical review of the manuscript, andMelinda Rance for preparation of the graphs.

REFERENCES

1. Parfitt AM, Mathews CH, Villanueva AR, Kleerekoper M,Frame B, Rao DS 1983 Relationships between surface, vol-ume, and thickness of iliac trabecular bone in aging and inosteoporosis. Implications for the microanatomic and cellularmechanisms of bone loss. J Clin Invest 72:1396–1409.

2. NIH Consensus Development Panel 2001 Osteoporosis pre-vention, diagnosis, and therapy. JAMA 285:785-795.

3. Recker RR, Barger-Lux MJ 2004 The elusive concept of bonequality. Curr Osteoporos Rep 2:97–100.

4. Riggs BL, Parfitt AM 2005 Drugs used to treat osteoporosis:The critical need for a uniform nomenclature based on theiraction on bone remodeling. J Bone Miner Res 20:177–184.

5. Reeve J, Bradbeer JN, Arlot M, Davies UM, Green JR, Hamp-ton L, Edouard C, Hesp R, Hulme P, Ashby JP 1991 hPTH1-34 treatment of osteoporosis with added hormone replace-ment therapy: Biochemical, kinetic and histological responses.Osteoporos Int 1:162–170.

6. Hodsman AB, Fraher LJ, Ostbye T, Adachi JD, Steer BM1993 An evaluation of several biochemical markers for boneformation and resorption in a protocol utilizing cyclical para-thyroid hormone and calcitonin therapy for osteoporosis. JClin Invest 91:1138–1148.

7. Hodsman AB, Fraher LJ, Watson PH, Ostbye T, Stitt LW,Adachi JD, Taves DH, Drost D 1997 A randomized controlledtrial to compare the efficacy of cyclical parathyroid hormoneversus cyclical parathyroid hormone and sequential calcitoninto improve bone mass in postmenopausal women with osteo-porosis. J Clin Endocrinol Metab 82:620–628.

8. Eastell R, Delmas PD, Hodgson SF, Eriksen EF, Mann KG,Riggs BL 1988 Bone formation rate in older normal women:Concurrent assessment with bone histomorphometry, calciumkinetics, and biochemical markers. J Clin Endocrinol Metab67:741–748.

MA ET AL.862

9. Dobnig H, Sipos A, Jiang Y, Fahrleitner-Pammer A, Ste-MarieLG, Gallagher JC, Pavo I, Wang J, Eriksen EF 2005 Earlychanges in biochemical markers of bone formation correlatewith improvements in bone structure during teriparatidetherapy. J Clin Endocrinol Metab 90:3970–3977.

10. Arlot M, Meunier PJ, Boivin G, Haddock L, Tamayo J, Cor-rea-Rotter R, Jasqui S, Donley DW, Dalsky GP, Martin JS,Eriksen EF 2005 Differential effects of teriparatide and alen-dronate on bone remodeling in postmenopausal women as-sessed by histomorphometric parameters. J Bone Miner Res20:1244–1253.

11. Frost HM 1990 Skeletal structural adaptations to mechanicalusage (SATMU): 1. Redefining Wolff’s law: The bone model-ing problem. Anat Rec 226:403–413.

12. Kobayashi S, Takahashi HE, Ito A, Saito N, Nawata M, Horiu-chi H, Ohta H, Ito A, Iorio R, Yamamoto N, Takaoka K 2003Trabecular minimodeling in human iliac bone. Bone 32:163–169.

13. Chavassieux PM, Arlot ME, Reda C, Wei L, Yates AJ, Meu-nier PJ 1997 Histomorphometric assessment of the long-termeffects of alendronate on bone quality and remodeling in pa-tients with osteoporosis. J Clin Invest 100:1475–1480.

14. Eriksen EF, Melsen F, Sod E, Barton I, Chines A 2002 Effectsof long-term risedronate on bone quality and bone turnover inwomen with postmenopausal osteoporosis. Bone 31:620–625.

15. Neer RM, Arnaud CD, Zanchetta JR, Prince R, Gaich GA,Reginster JY, Hodsman AB, Eriksen EF, Ish-Shalom S,Genant HK, Wang O, Mitlak BH 2001 Effect of parathyroidhormone (1-34) on fractures and bone mineral density in post-menopausal women with osteoporosis. N Engl J Med344:1434–1441.

16. Orwoll E, Scheele WH, Paul S, Adami S, Syversen U, Diez-Perez A, Kaufman JM, Clancy AD, Gaich G 2003 The effect ofteriparatide [human parathyroid hormone (1-34)] therapy onbone mineral density in men with osteoporosis. J Bone MinerRes 18:9–17.

17. Jiang Y, Zhao JJ, Mitlak BH, Wang O, Genant HK, EriksenEF 2003 Teriparatide [recombinant human parathyroid hor-mone (1-34)] improves both cortical and cancellous bone struc-ture. J Bone Miner Res 18:1932–1941.

18. Dempster DW, Cosman F, Kurland ES, Zhou H, Nieves J,Woelfert L, Shane E, Plavetic K, Muller R, Bilezikian J, Lind-say R 2001 Effects of daily treatment with parathyroid hor-mone on bone microarchitecture and turnover in patients withosteoporosis: A paired biopsy study. J Bone Miner Res16:1846–1853.

19. Seeman E, Delmas PD 2001 Reconstructing the skeleton withintermittent parathyroid hormone. Trends Endocrinol Metab12:281–283.

20. Reeve J, Meunier PJ, Parsons JA, Bernat M, Bijvoet OL,Courpron P, Edouard C, Klenerman L, Neer RM, Renier JC,Slovik D, Vismans FJ, Potts JT Jr 1980 Anabolic effect ofhuman parathyroid hormone fragment on trabecular bone ininvolutional osteoporosis: A multicentre trial. BMJ 280:1340–1344.

21. Bradbeer JN, Arlot ME, Meunier PJ, Reeve J 1992 Treatmentof osteoporosis with parathyroid peptide (hPTH 1-34) and oes-trogen: Increase in volumetric density of iliac cancellous bonemay depend on reduced trabecular spacing as well as increasedthickness of packets of newly formed bone. Clin Endocrinol(Oxf) 37:282–289.

22. Hodsman AB, Steer BM 1993 Early histomorphometricchanges in response to parathyroid hormone therapy in osteo-porosis: Evidence for de novo bone formation on quiescentcancellous surfaces. Bone 14:523–527.

23. Dempster DW, Zhou H, Cosman F, Nieves J, Adachi JD, Fra-her LJ, Watson PH, Lindsay R, Hodsman AB 2001 PTH treat-ment directly stimulates bone formation in cancellous and cor-tical bone in humans. J Bone Miner Res 16:S1;S179.

24. Yao W, Jee WS, Zhou H, Lu J, Cui L, Setterberg R, Liang T,Ma Y 1999 Anabolic effect of prostaglandin E2 on corticalbone of aged male rats comes mainly from modeling-dependent bone gain. Bone 25:697–702.

25. Dobnig H, Turner RT 1995 Evidence that intermittent treat-ment with parathyroid hormone increases bone formation inadult rats by activation of bone lining cells. Endocrinology136:3632–3638.

26. Hock JM, Hummert JR, Boyce R, Fonseca J, Raisz LG 1989Resorption is not essential for the stimulation of bone growthby hPTH-(1-34) in rats in vivo. J Bone Miner Res 4:449–458.

27. Lindsay R, Cosman F, Zhou H, Bostrom MP, Shen VW, CruzJD, Nieves JW, Dempster DW 2006 A novel tetracycline la-beling scheduling for longitudinal evaluation of the short-termeffects of anabolic therapy with a single iliac crest bone biopsy:Early actions of teriparatide. J Bone Miner Res 21:366–373.

28. Lindsay R, Nieves J, Formica C, Henneman E, Woelfert L,Shen V, Dempster D, Cosman F 1997 Randomised controlledstudy of effect of parathyroid hormone on vertebral-bone massand fracture incidence among postmenopausal women on oes-trogen with osteoporosis. Lancet 350:550–555.

29. Erben RG 1996 Trabecular and endocortical bone surfaces inthe rat: Modeling or remodeling? Anat Rec 246:39–46.

30. Canalis E 1995 Growth hormone, skeletal growth factors andosteoporosis. Endocr Pract 1:39–43.

31. Yakar S, Rosen CJ, Beamer WG, Ackert-Bicknell CL, Wu Y,Liu JL, Ooi GT, Setser J, Frystyk J, Boisclair YR, LeRoith D2002 Circulating levels of IGF-1 directly regulate bone growthand density. J Clin Invest 110:771–781.

32. Yakar S, Rosen CJ 2003 From mouse to man: Redefining therole of insulin-like growth factor-I in the acquisition of bonemass. Exp Biol Med (Maywood) 228:245–252.

33. Canalis E, McCarthy TL, Centrella M 1990 Differential effectsof continuous and transient treatment with parathyroid hor-mone related peptide (PTHrp) on bone collagen synthesis. En-docrinology 126:1806–1812.

34. Kasukawa Y, Miyakoshi N, Mohan S 2004 The anabolic effectsof GH/IGF system on bone. Curr Pharm Des 10:2577–2592.

35. Boonen S, Mohan S, Dequeker J, Aerssens J, VanderschuerenD, Verbeke G, Broos P, Bouillon R, Baylink DJ 1999 Down-regulation of the serum stimulatory components of the insulin-like growth factor (IGF) system (IGF-I, IGF-II, IGF bindingprotein [BP]-3, and IGFBP-5) in age-related (type II) femoralneck osteoporosis. J Bone Miner Res 14:2150–2158.

36. Vahle JL, Sato M, Long GG, Young JK, Francis PC, Engel-hardt JA, Westmore M, Ma L, Nold JB 2002 Skeletal changesin rats given daily subcutaneous injections of recombinant hu-man parathyroid hormone(1-34) for 2 years and relevance tohuman safety. Toxicol Pathol 30:312–321.

37. Vahle JL, Long GG, Sandusky G, Westmore M, Ma YL, SatoM 2004 Bone neoplasms in F344 rats given teriparatide[rhPTH(1-34)] are dependent on duration of treatment anddose. Toxicol Pathol 32:426–438.

38. Eriksen EF 1986 Normal and pathological remodeling of hu-man trabecular bone: Three dimensional reconstruction of theremodeling sequence in normals and in metabolic bone dis-ease. Endocr Rev 7:379–408.

39. Parfitt AM, Drezner MK, Glorieux FH, Kanis JA, Malluche H,Meunier PJ, Ott SM, Recker RR 1987 Bone histomorphom-etry: Standardization of nomenclature, symbols, and units. Re-port of the ASBMR Histomorphometry Nomenclature Com-mittee. J Bone Miner Res 2:595–610.

40. Ma Y, Jee WS, Chen Y, Gasser J, Ke HZ, Li XJ, Kimmel DB1995 Partial maintenance of extra cancellous bone mass byantiresorptive agents after discontinuation of human parathy-roid hormone (1-38) in right hindlimb immobilized rats. J BoneMiner Res 10:1726–1734.

41. Erben Reinhold G 1997 Embedding of bone samples in methy-methacrylate: An improved method suitable for bone histo-morphometry, histochemistry, and immunohistochemistry. JHistochem Cytochem 45:307–313.

42. Hodsman AB, Kisiel M, Adachi JD, Fraher LJ, Watson PH2000 Histomorphometric evidence for increased bone turnoverwithout change in cortical thickness or porosity after 2 years ofcyclical hPTH(1-34) therapy in women with severe osteoporo-sis. Bone 27:311–318.

TERIPARATIDE-INDUCED MODELING BONE FORMATION 863

43. McClung MR, San Martin J, Miller PD, Civitelli R, Bandeira F,Omizo M, Donley DW, Dalsky GP, Eriksen EF 2005 Teripa-ratide and alendronate increase bone mass by opposite effectson bone remodeling. Arch Intern Med 165:1762–1768.

44. Misof BM, Roschger P, Cosman F, Kurland ES, Tesch W,Messmer P, Dempster DW, Nieves J, Shane E, Fratzl P,Klaushofer K, Bilezikian J, Lindsay R 2003 Effects of inter-mittent parathyroid hormone administration on bone mineral-ization density in iliac crest biopsies from patients with osteo-porosis: A paired study before and after treatment. J ClinEndocrinol Metab 88:1150–1156.

45. Frost HM 1986 Intermediary Organization of the Skeleton.CRC Press, Boca Raton, FL, USA.

46. Eriksen EF, Mosekilde L, Melsen F 1986 Kinetics of trabecularbone resorption and formation in hypothyroidism: Evidencefor a positive balance per remodeling cycle. Bone 7:101–108.

47. Mohan S, Baylink DJ 1991 Evidence that the inhibition ofTE85 human bone cell proliferation by agents which stimulatecAMP production may in part be mediated by changes in theIGF-II regulatory system. Growth Regul 1:110–118.

48. Pepene CE, Seck T, Diel I, Minne HW, Ziegler R, PfeilschifterJ 2004 Influence of fluor salts, hormone replacement therapyand calcitonin on the concentration of insulin-like growth fac-tor (IGF)-I, IGF-II and transforming growth factor-beta 1 inhuman iliac crest bone matrix from patients with primary os-teoporosis. Eur J Endocrinol 150:81–91.

49. Turner CH 2002 Biomechanics of bone: Determinants of skel-etal fragility and bone quality. Osteoporos Int 13:97–104.

50. Martin RB 2002 Size, structure and gender: Lessons about frac-ture risk. J Musculoskelet Neuronal Interact 2:209–211.

51. Seck T, Scheppach B, Scharla S, Diel I, Blum WF, Bismar H,Schmid G, Krempien B, Ziegler R, Pfeilschifter J 1998 Con-centration of insulin-like growth factor (IGF)-I and -II in iliaccrest bone matrix from pre- and postmenopausal women: Re-lationship to age, menopause, bone turnover, bone volume,and circulating IGFs. J Clin Endocrinol Metab 83:2331–2337.

52. Ishibe M, Ishibashi T, Kaneda K, Koda T, Rosier RN, Puzas JE1998 Stimulation of bone formation in vivo by insulin-likegrowth factor-II in rats. Calcif Tissue Int 63:36–38.

Address reprint requests to:Yanfei L Ma, MD

Lilly Research LaboratoriesDC0403

Eli Lilly and CompanyIndianapolis, IN 46285, USAE-mail: [email protected]

Received in original form June 13, 2005; revised form February 27,2006; accepted March 28, 2006.

MA ET AL.864