A Novel Tetracycline Labeling Schedule for Longitudinal Evaluation of the Short-Term Effects of Anabolic Therapy With a Single Iliac Crest Bone Biopsy: Early Actions of Teriparatide

Authors

  • Robert Lindsay MD, PhD,

    Corresponding author
    1. Clinical Research Center, New York State Department of Health, Helen Hayes Hospital, West Haverstraw, New York, USA
    2. Department of Medicine, Columbia University College of Physicians and Surgeons, New York, New York, USA
    • Clinical Research Center, Helen Hayes Hospital, Route 9W, West Haverstraw, NY 10993, USA
    Search for more papers by this author
  • Felicia Cosman,

    1. Clinical Research Center, New York State Department of Health, Helen Hayes Hospital, West Haverstraw, New York, USA
    2. Department of Medicine, Columbia University College of Physicians and Surgeons, New York, New York, USA
    Search for more papers by this author
  • Hua Zhou,

    1. Regional Bone Center, New York State Department of Health, Helen Hayes Hospital, West Haverstraw, New York, USA
    Search for more papers by this author
  • Mathias P Bostrom,

    1. Hospital for Special Surgery, New York, New York, USA
    Search for more papers by this author
  • Victor W Shen,

    1. Skeletech, Inc., Bothel, Washington, USA
    Search for more papers by this author
  • JoAnn D Cruz,

    1. Clinical Research Center, New York State Department of Health, Helen Hayes Hospital, West Haverstraw, New York, USA
    Search for more papers by this author
  • Jeri W Nieves,

    1. Clinical Research Center, New York State Department of Health, Helen Hayes Hospital, West Haverstraw, New York, USA
    2. Department of Epidemiology, Columbia University College of Physicians and Surgeons, New York, New York, USA
    Search for more papers by this author
  • David W Dempster

    1. Department of Medicine, Columbia University College of Physicians and Surgeons, New York, New York, USA
    2. Department of Pathology, Columbia University College of Physicians and Surgeons, New York, New York, USA
    Search for more papers by this author

  • The authors state that they have no conflicts of interest.

Abstract

We describe a quadruple tetracycline labeling method that allows longitudinal assessment of short-term changes in bone formation in a single biopsy. We show that 1 month of hPTH(1-34) treatment extends the bone-forming surface, increases mineral apposition rate, and initiates modeling-based formation.

Introduction: Iliac crest biopsy, with histomorphometric evaluation, provides important information about cellular activity in bone. However, to obtain longitudinal information, repeat biopsies must be performed. In this study, we show the capability to obtain short-term longitudinal information on bone formation in a single biopsy using a novel, quadruple labeling technique.

Materials and Methods: Two tetracycline labels were administered using a standard 3 days on, 12 days off, 3 days on format. Four weeks later, the tetracycline labeling was repeated using the same schedule but with a different tetracycline that can be distinguished from the first by its color under fluorescent light. Iliac crest biopsies were performed 1 week later and prepared undecalcified for histomorphometry. Indices of bone formation 1 month apart were measured and calculated using the two sets of labels. We used this method to investigate the early effects of teriparatide [hPTH(1-34)] treatment on bone formation. The results were compared with those from a group of control subjects who were quadruple-labeled, but did not receive hPTH(1-34).

Results: Treatment with hPTH(1-34) dramatically stimulated bone formation on cancellous and endocortical surfaces. This was achieved by both an increase in the linear rate of matrix apposition and extension of the bone-forming surface. New bone was deposited on previously quiescent surfaces (i.e., modeling-based formation), but a proportion of this could occur by encroachment from adjacent resorption cavities.

Conclusions: A single transiliac crest bone biopsy, after sequential administration of two sets of tetracycline labels is a useful approach to study the short-term effects of anabolic agents on human bone. One month of hPTH(1-34) treatment extends the bone-forming surface, increases mineral apposition rate, and initiates modeling-based formation.

INTRODUCTION

THE USE OF bone biopsies from the iliac crest is a powerful technique for examination of bone remodeling and its cellular mechanisms.(1) However, each biopsy gives a window on bone behavior at a single time-point and to obtain longitudinal data requires a second biopsy. The second biopsy must be obtained from the contralateral iliac crest, increasing the variability of the measurements.(2-7) The limitations of this approach are well known. Many individuals will not return for the second biopsy, and for practical reasons, the interbiopsy interval is usually not less than 6 months. Shorter-term perturbations in remodeling are difficult to study in this fashion.

The object of this study was to determine if short-term, longitudinal information on bone formation could be obtained from a single iliac crest biopsy using four tetracycline labels (two sets of double labels). To study the capability to observe perturbations in bone formation, after completion of the first set of labels, one-half of the subjects were randomly assigned to receive hPTH(1-34) 25 μg/day by daily subcutaneous injection. This treatment was continued during administration of the second set of labels and until the day of biopsy. The other subjects served as controls and received no hPTH(1-34).

MATERIALS AND METHODS

Subjects

Postmenopausal women (n = 21) with low bone mass or osteoporosis were recruited for the study through our BMD screening center and osteoporosis clinic. Women with secondary causes of bone loss (current steroid treatment, hyperparathyroidism, hypercalcemia, vitamin D deficiency) were excluded, but women on osteoporosis medication such as alendronate, risedronate, estrogen, or raloxifene were allowed to participate. All women were advised to maintain total calcium intakes of at least 1200 mg/day and vitamin D intakes of at least 600 IU/day. The study was approved by the Institutional Review Board of Helen Hayes Hospital, and all women gave written informed consent.

Protocol

To prepare for biopsy, subjects were given two sets of tetracycline labels in the following way. The first set of labels was given to all subjects in standard format: 3 days of demeclocycline HCl (150 mg, four times per day), a 12-day intermission, and 3 more days of demeclocycline (same dose). One-half of the subjects were randomly assigned to receive hPTH(1-34) 25 μg/day by daily subcutaneous injection after the first set of labels was administered. This treatment was continued during administration of the second set of labels and until the day of biopsy. The other subjects served as controls and received no hPTH(1-34). Four weeks after completion of the first set of labels, a second set of double labels was administered following the same schedule, but this time using tetracycline HCl (250 mg, four times per day for 3 days on, 12 days off, 3 days on). Five of the 11 women in the hPTH(1-34)-treated group were on antiresorptive therapy (alendronate or hormone therapy); 3 of the 10 women in the control group were on antiresorptive therapy (alendronate, risedronate, or hormone therapy). The antiresorptive therapy was continued during the treatment period in both groups.

Standard preoperative laboratory tests (complete blood count [CBC], glucose, electrolyte panel, prothrombin time) as well as an electrocardiogram were obtained 1 week before biopsy. Standard transiliac crest bone biopsies were performed 6 days after the last day of tetracycline administration following standard procedures.(1) Biopsies were embedded in methacrylate and prepared in a standard fashion for histomorphometric analysis.(1) Serial sections from each patient were examined under UV light for evaluation of tetracycline labels and under polarized light after staining with methylene blue for evaluation of reversal and cement lines. In addition to the estimation of the dynamic bone formation variables as previously described,(1) we identified different types of formation, based on pre-established criteria (Fig. 1). Where we identified bone formation occurring over a scalloped reversal line, we interpreted this as bone formation after resorption, as expected in a traditional remodeling unit. We called this remodeling-based formation. We also identified bone formation over smooth cement lines and designated this as modeling-based formation. All other histomorphometric variables were defined and expressed according to the recommendations of the ASBMR nomenclature committee.(8) The individual reading the biopsies (HZ) was blinded to the treatment assignment of the patients

Figure FIG. 1..

Morphological characteristics of remodeling-based bone forming units (top) and modeling-based bone forming units (bottom). Double labels overlying a scalloped cement line were considered to represent remodeling-based formation, whereas those overlying a smooth cement line were considered to represent modeling-based formation.20

Statistical analysis

Variables were expressed as means ± SE. Statistical analysis was performed using the NCSS Statistical System (NCSS, Kaysville, UT, USA). The differences in baseline demographics, structural variables, and static and dynamic indices of bone formation between hPTH(1-34)-treated and control groups were analyzed using two-sample t-tests. Paired t-tests were performed to compare the differences in the dynamic indices between the two time-points derived from the two sets of double labels. A p value of <0.05 was considered to be significant.

RESULTS

The baseline demographic details of the population are shown in Table 1. There were no significant differences between groups. Subjects were on average ∼60 years of age and 7-10 years postmenopausal. Height, weight, and BMD did not differ between the groups. Structural indices for both trabecular and cortical bone are shown in Table 2. Cancellous bone volume was low in both groups, compatible with their low bone mass status. Trabecular width, number, and spacing were similar in both groups. Cortical width was significantly lower in the control group than in the hPTH(1-34)-treated group. There were no significant differences in baseline bone formation variables derived from the first set of labels between the control and hPTH(1-34)-treated subjects (Table 3).

Table Table 1.. Baseline Demographics of the hPTH(1-34)-Treated Patients and Controls
original image
Table Table 2.. Bone Structure Variables in Control Subjects and Subjects Treated With hPTH(1-34) (Mean ± SE)
original image
Table Table 3.. Bone Formation Variables From Two Sets of Tetracycline Labels in Control Subjects and Subjects Treated With hPTH(1-34) (Mean ± SE)
original image

Static indices of bone remodeling in hPTH(1-34)-treated subjects versus controls

Cross-sectional evaluation of static indices of bone remodeling in cancellous bone showed the stimulatory effect of hPTH(1-34). The percent surface covered by osteoid was 5.4 ± 1.3% in control subjects and 10.8 ± 1.8% in hPTH(1-34)-treated subjects (p < 0.05). Osteoid thickness was slightly, but not significantly, higher (3.15 versus 2.53 lamellae), and the proportion of osteoid occupied by active osteoblasts was higher (p < 0.01) in hPTH(1-34)-treated versus control subjects (10.6 ± 1.28% versus 3.27 ± 0.5%). In contrast, eroded perimeter was not higher in hPTH(1-34)-treated subjects at this early time-point after hPTH(1-34) administration (1.39 ± 0.18% versus 2.24 ± 0.41% in controls) with 0.6% and 0.9% of eroded perimeter covered with active osteoclasts in the hPTH(1-34) and control groups, respectively. There was also no difference between groups in the average length of the eroded perimeter underlying the second set of labels [0.40 ± 0.08 in hPTH(1-34) versus 0.52 ± 0.05 mm in controls].

Dynamic indices of bone formation in first versus second set of labels

Figure 2 shows typical quadruple labels in trabecular bone. The upper and lower panels show two sets of labels in an hPTH(1-34)-treated and a control subject, respectively. The data from control and hPTH(1-34)-treated subjects are summarized in Table 3 for three skeletal envelopes (cancellous, endocortical, and intracortical). In control subjects, within both cancellous and endocortical bone, mean mineralizing perimeter, mineral apposition rate, and bone formation rate were all constant from the first to the second set of labels, with no significant differences between the two. In contrast, in the hPTH(1-34)-treated group, there were significant increases in cancellous and endocortical mineralizing perimeter, mineral apposition rate, and bone formation rate derived from the second set of labels [during hPTH(1-34) treatment] versus the first set of labels [pre-PTH(1-34) treatment]. For example, bone formation rate was approximately doubled in the endocortex and tripled in cancellous bone after hPTH(1-34) administration. Within cortical bone, there were no significant differences in the measured variables between the first and second set of labels in either hPTH(1-34) or control groups (Table 3).

Figure FIG. 2..

Quadruple labels in hPTH(1-34)-treated subject (top) and a control subject (bottom).20

At all sites where quadruple labels were present, the average length of each second double label was significantly higher than that of the first double label in the hPTH(1-34) group in both cancellous and endocortical envelopes. Conversely, in the control group, the average length of each second double label was slightly, but not significantly, lower than the first (Figs. 2 and 3).

Figure FIG. 3..

Average length of individual double labels at quadruple-labeled sites in (A) hPTH(1-34)-treated patients and (B) controls in cancellous, endocortical, and intracortical envelopes.20

We assigned posttreatment double labels to either remodeling-based formation or modeling-based formation, depending on whether the underlying reversal line was scalloped or smooth (Figs. 1 and 4). In the control subjects, all formation was remodeling-based, whereas in the hPTH(1-34)-treated patients, ∼70% was remodeling-based and 30% was modeling-based (Table 4). In the hPTH(1-34)-treated subjects, we frequently observed remodeling units in which the second double label extended beyond the limits of the scalloped reversal line onto the adjacent, previously unresorbed, surface (Fig. 5). Such extended remodeling units constituted 49% and 64% of all remodeling units in the cancellous and endocortical envelopes, respectively. They were observed in 9 of the 10 hPTH(1-34)-treated subjects, but only one such unit was seen in only 1 of the 9 control subjects. Most of the modeling-based formation in the hPTH(1-34)-treated subjects occurred in such extended remodeling units. In 4 of 10 hPTH(1-34) subjects, we did observe modeling-based formation that was not associated with adjacent, eroded surface. This was never seen in the control subjects.

Figure FIG. 4..

Examples of remodeling-based formation (top two panels) and modeling-based formation (bottom two panels). The right panels show the fluorescent image of the double labels corresponding to the toluidine blue sections in the left panels. Note the scalloped cement line in the top left panel (white arrows) and the smooth cement line in the bottom left panel (yellow arrows).20

Figure FIG. 5..

New bone formation under the influence of hPTH(1-34) treatment. The top panel shows the fluorescent image of the double labels corresponding to the toluidine blue sections in the bottom panel. Note that the second set of labels extends to the right beyond the margin of the scalloped reversal line (between the two arrows).20

Table Table 4.. Remodeling- and Modeling-Based Formation in hPTH(1-34)-Treated Patients and Controls
original image

DISCUSSION

We have used a novel, quadruple tetracycline labeling schedule to obtain longitudinal information on bone formation from a single iliac crest biopsy. Using two sets of double tetracycline labels with two tetracyclines, which differ in their color under fluorescent light, we were able to distinguish the first and second sets of tetracycline labels, even when only single labels were present. Using standard histomorphometry, mineral apposition rate and mineralizing perimeter can be measured for the two separate labeling periods, and bone formation rates can be calculated. The technique has two key advantages over paired biopsies. First, only one biopsy is required. Second, each patient serves as his or her own pretreatment control, eliminating problems caused by the large intersite variability in histomorphometric variables.(2-7) As a result, with only a small number of biopsies in the hPTH(1-34) group (n = 10), we were able to show highly significant differences in dynamic variables after a 1-month treatment period. The data from the control subjects indicated that, without intervention, all bone formation variables were stable over the 1-month period. Triple tetracycline labels have previously been used to test the hypothesis that pauses occur during the bone formation period.(9) This is the first study in which a multiple labeling approach has been used to assess the short-term effects of treatment with a bone-active agent.

The cellular mechanisms underlying the effects of anabolic agents on bone remain poorly understood. Here we used the quadruple labeling technique to gain new information on the early effects of daily injections of hPTH(1-34). hPTH(1-34) treatment dramatically stimulated bone formation on both cancellous and endocortical surfaces. The increase in bone formation was achieved by two mechanisms: an increase in the linear rate of mineral apposition and an increase in extent of bone-forming surface. The increase in mineral apposition rate at sites in which two double labels were present indicates that hPTH(1-34) is able to stimulate formation in remodeling units that were active before initiation of treatment. This could be achieved by stimulation of the production rate of preexisting osteoblasts, by enhanced recruitment of osteoblasts into preexisting bone-forming units, and/or by an increase in osteoblast life span.(10)

Another interesting early effect of hPTH(1-34) was the increase in mineralizing perimeter in the cancellous and endocortical envelopes. This was primarily because of an increase in the average length of individual labels, and this, again, occurred in remodeling units that were active before initiation of therapy (i.e., those with quadruple labels). Because bone formation occurs in three dimensions, this observation suggests that hPTH(1-34) is capable of rapidly and dramatically increasing the surface area of individual bone-forming units. The stimulation of bone formation on cancellous and endocortical surfaces provides the cellular basis for increased wall thickness of cancellous and endocortical bone packets,(11-13) and this in turn provides the structural basis for increased trabecular and cortical thickness.(11,13,14) This is supported by a recent observation that changes in biochemical markers of bone formation 1 month after initiation of hPTH(1-34) treatment correlate with changes in bone structure and BMD at 22 months.(15,16) There have been reports that prior or ongoing treatment with bisphosphonates attenuates the anabolic action of hPTH(1-34) and hPTH(1-84).(17-19) Whereas some of the patients in our study were taking bisphosphonates before and during the treatment period [two in the hPTH(1-34)-treated group and two in the control group], the number of subjects was too small to address this issue. Similarly, the small number of patients precluded a meaningful subanalysis of the effects of antiresorptive therapy (bisphosphonates or hormone replacement therapy) on the response to hPTH(1-34).

The quadruple labeling technique described here is not without limitations, which include the need to administer the second set of labels within a reasonable time frame to maintain the capability to obtain reliable information from the first set of labels. Because we have not performed a study with variable time periods between the two sets of labels, we do not know exactly how long that period might be. Furthermore, this time period would differ for the treatment agent under study. For example, the greater the extent to which an agent increases bone turnover and the longer the interlabel interval, the greater is the likelihood of a substantial portion of the first set of labels being resorbed, thus confounding the analysis. Moreover, it should be noted that this technique cannot be used to assess the steady-state effects of drugs on bone. Its primary use will be to explore the short-term effects of anabolic agents and, perhaps, of agents that display mixed antiresorptive and anabolic activity. The quadruple labeling technique allows assessment of changes in dynamic variables, but not changes in static remodeling variables, such as resorption indices or structural variables, which can only be assessed at one time-point. The technique is, therefore, not intended to replace traditional, paired, or cross-sectional biopsy studies, which are often used by the pharmaceutical industry to assess drug safety and can provide information on many variables in addition to those related to bone formation. However, our technique does allow comparison of values for static and structural variables in subjects receiving treatment with those in parallel control biopsies or with age- and sex-matched normative data. For example, in this study, whereas we showed a significant increase in bone formation variables after 1 month of treatment with hPTH(1-34), there was no difference in the eroded perimeter or the osteoclast perimeter compared with controls. On the other hand, static indices of bone formation, such as the osteoblast and osteoid perimeters, were higher than those in controls, consistent with the longitudinal increases in dynamic bone formation variables.

As in previous studies,(20-22) we assigned new bone formation as being modeling-based or remodeling-based, depending on whether the underlying cement line was smooth or scalloped. The assumption that a smooth cement line implies modeling (i.e., formation without prior resorption), has never been, and probably cannot be, experimentally confirmed. There are other possible explanations. For example, osteoclasts could erode down to a lamellar plate in the bone and resorb along it without leaving a scalloped cement line. Whatever the mechanism, packets with smooth cement lines, are less common than those with scalloped cement lines under normal circumstances. In this study, we only observed smooth cement lines in one of nine control subjects. Takahashi et al.(20) reported that 100% of 6000 cement lines in cortical bone and 97% of 5400 cement lines in cancellous bone from 157 normal adults were scalloped. In a study of iliac crest biopsy from 34 patients undergoing total hip arthroplasty, 38% of the individuals did not have packets with smooth cement lines, and they were rarely present in 53% of the individuals.(22)

It was postulated a number of years ago(23) that daily PTH treatment can induce so-called de novo bone formation on previously quiescent surfaces, or what we have termed here modeling-based bone formation. One potential mechanism proposed to account for this phenomenon is resumption of bone formation by bone-lining cells.(24) In this study, we assessed the proportions of remodeling- and modeling-based bone formation in controls and after 1 month of hPTH(1-34) treatment. During hPTH(1-34) treatment, the majority (70%) of new bone formation was occurring over scalloped reversal lines and, therefore, was remodeling-based. Because the eroded perimeter was not higher than in the controls, we conclude that the bulk of new bone formation during this early treatment period occurs in remodeling units in which bone resorption was already ongoing or completed at the start of treatment. We also observed a substantial proportion (30%) of modeling-based formation during hPTH(1-34) treatment. This could represent the initiation of new bone formation that is both spatially and temporally unrelated to prior resorption. We believe, however, that much of the modeling-based formation is the result of extension of bone formation to quiescent surfaces adjacent to the original resorption cavity, as originally hypothesized by Parfitt.(25) We have shown for the first time that this does indeed occur (Fig. 5). Furthermore, we suggest that, even if new formation is observed on quiescent surfaces, this could still represent “spill-over” formation from a resorption cavity that is outside the section plane (Fig. 6). Our data do not, however, preclude the possibility that PTH is capable of stimulating new bone formation that is completely unrelated to prior resorption, both spatially and temporally. Initiation of new bone formation on quiescent surfaces surrounding the original resorption cavity could occur by osteoblasts migrating out of the cavity to annex the neighboring territory, or, conceivably, by the activation of adjacent lining cells by osteoblast-derived, paracrine factors, such as IGF-I.

Figure FIG. 6..

Proposed mechanism whereby, under the influence of hPTH(1-34) treatment, bone formation extends beyond the limits of the remodeling unit to lay down new bone on adjacent, previously unresorbed surface. (A) The osteoblasts may refill the resorption cavity and “spill over” onto the adjacent territory. An example of this is shown in Fig. 5. However, if the plane of section does not include the scalloped reversal line (e.g., section orthogonal to the plane of the paper along the line XY as shown in B), new bone formation may erroneously be interpreted to be occurring on smooth bone surfaces (arrow) completely unrelated to a resorptive event.20

We have previously reported evidence of anabolic action of continuously elevated, endogenous PTH on cancellous bone in postmenopausal women with mild primary hyperparathyroidism (PHPT).(26) We therefore thought it would be of interest to assess the frequency of modeling-based formation in that situation. We compared data obtained using only the second set of labels from the hPTH(1-34)-treated subjects in this study with unpublished data from a subset of 11 age-matched women with PHPT who had received traditional double labels and who were previously described in detail.(23) In cancellous bone, modeling-based formation was observed in extended remodeling units in 3 of 11 (27%) of the PHPT patients compared with 9 of 10 (90%) of the hPTH(1-34)-treated patients. They constituted 2% of all forming units in the patients with PHPT compared with 49% in the hPTH(1-34)-treated patients. In the PHPT patients, there was no modeling-based formation that was not associated with adjacent resorption, whereas this was seen in 40% of the hPTH(1-34)-treated patients. These findings suggest that modeling-based formation is much more common in patients treated with daily injections of hPTH(1-34) than in patients with chronically elevated, endogenous PTH.

In conclusion, a single transiliac crest bone biopsy, after sequential administration of two sets of tetracycline labels, is a useful approach to study the short-term effects of anabolic treatments on human bone. One month of treatment with hPTH(1-34) dramatically stimulates bone formation on cancellous and endocortical surfaces. This is achieved by both an increase in the linear rate of matrix apposition and extension of the bone-forming surface. New bone is deposited on previously quiescent surfaces, but a proportion of this may occur by encroachment from adjacent remodeling units.

Acknowledgements

This study was supported by National Institutes of Health Grant AR39101, an unrestricted research grant from Bristol-Myers Squibb, and the Helen Hayes Hospital Foundation.

Ancillary