Osteocyte density in aging subjects is enhanced in bone adjacent to remodeling haversian systems

Bone Vol. 30, No. 6 June 2002:859 – 865

Osteocyte Density in Aging Subjects Is Enhanced in Bone Adjacent to Remodeling Haversian Systems J. POWER,1 N. LOVERIDGE,1 N. RUSHTON,2 M. PARKER,3 and J. REEVE1 1 3

Bone Research Group (MRC), Department of Medicine, and 2Orthopaedic Research Unit, University of Cambridge Clinical School, Cambridge, UK Peterborough District Hospital, Peterborough, UK

Introduction The osteocyte is a candidate regulatory cell for bone remodeling. Previously, we demonstrated that there is a substantial (approximately 50%) loss of osteocytes from their lacunae in the cortex of the elderly femoral neck. Higher occupancy was evident in tissue exhibiting high remodeling and high porosity. The present study examines the distribution of osteocytes within individual osteonal systems at differing stages of the remodeling cycle. In 22 subjects, lacunar density, osteocyte density, and their quotient, the percent lacunar occupancy, was assessed up to a distance of 65 ␮m from the canal surface in six quiescent, resorbing, and forming osteons. In both forming (p ⴝ 0.024) and resorbing (p ⴝ 0.034) osteons, osteocyte densities were significantly higher in cases of hip fracture than controls. However, there were no significant between-group differences in lacunar occupancy. In both cases and controls, osteocyte density (p < 0.0001; mean difference ⴞSEM: 157 ⴞ 34/mm2) and lacunar occupancy (p ⴝ 0.025; mean difference: 8.1 ⴞ 3.4%) were shown to be significantly higher in forming compared with quiescent osteons. Interestingly, resorbing systems also exhibited significantly elevated osteocyte density in both the fracture and the control group combined (mean difference 76 ⴞ 23/mm2; p ⴝ 0.003). Lacunar occupancy was also greater in resorbing compared with quiescent osteons (both groups combined: p ⴝ 0.022; mean difference: 5.7 ⴞ 2.3%). Elevated osteocyte density and lacunar occupancy in forming compared with quiescent systems was expected because of the likely effects of aging on quiescent osteons. However, the higher levels of these parameters in resorbing compared with quiescent systems was the opposite of what we expected and suggests that, in addition to their postulated mechanosensory role in the suppression of remodeling and bone loss, osteocytes might also contribute to processes initiating or maintaining bone resorption. (Bone 30:859 – 865; 2002) © 2002 by Elsevier Science Inc. All rights reserved.

The osteocyte represents the final differentiation step in the osteoblastic lineage and arises when bone-forming osteoblasts become encased within the calcified matrix. Osteocytes form, through their dendritic processes, an intercommunicating network between individual cells and the bone surface lining cells. Although the functional role of the osteocyte remains partly unresolved, a key role in the regulation of remodeling the skeletal architecture in response to mechanical load has been suggested.4,11,18 A variety of mechanisms for explaining this role have been suggested; for example, stretch-activated ion channels7,12 or shear stress from the flow of interstitial fluid,28 leading to the release of paracrine mediators or cell-cell communication. In addition to sensing mechanical strain, it has been proposed that osteocytes may also detect microdamage within bone3 and direct its removal through their apoptosis.20 The direction of remodeling though the generation and delivery of signals to cells populating the bone surface, the osteoblasts and lining cells, represents a possible key osteocytic function. The true nature of the mechanism(s) by which osteocytes might influence and/or regulate the activity and numbers of effector cells of bone remodeling remains controversial. Marotti and coworkers proposed that osteoblast differentiation to the osteocyte cell type is regulated via an inhibitory signal relayed from already formed osteocytes.15,16 This concept has recently been expanded by Martin to develop a hypothesis for the downregulation of remodeling by osteocytes.17 In this proposal, an osteocytic inhibitory signal acts on lining cells, located on quiescent bone surfaces, effectively preventing these cells from initiating a remodeling sequence. Regions of bone exhibiting high rates of remodeling might then be expected to correlate with relatively low numbers of viable osteocytes, because low osteocyte density would reduce the proposed functional brake to remodeling. Somewhat at variance with Martin’s concept is the proposal by Noble and Reeve that resorption is targeted to the matrix surrounding osteocytes undergoing apoptosis.20 According to this hypothesis, at times of stress, such as when osteocytes begin to undergo apoptosis, they may promote (rather than inhibit) remodeling. As a consequence, observed rates of osteocyte apoptosis would be expected to show a positive correlation with rates of remodeling. These investigators found that, within human infant and pathological bone, osteocyte apoptosis was positively associated with bone turnover,20 whereas a further study on iliac bone25 demonstrated that suppression of estrogen in women induced osteocyte apoptosis associated with reduced connectivity of cancellous bone.6 Furthermore, a recent study by Terai et al. implicated expression of

Key Words: Femoral neck; Osteocyte density; Osteocyte lacunae; Cortical bone; Bone remodeling; Osteonal systems.

Address for correspondence and reprints: Dr. J. Power, Department of Medicine, Addenbrooke’s Hospital (Box 157), Hills Road, Cambridge CB2 2QQ. E-mail: [email protected] © 2002 by Elsevier Science Inc. All rights reserved.

859

8756-3282/02/$22.00 PII S8756-3282(02)00731-7

860

J. Power et al. Osteocytes in quiescent and remodeling osteons

Bone Vol. 30, No. 6 June 2002:859 – 865

Table 1. Control subjects and their recorded causes of death Subject

Age (years)

Cause of death

1 2 3 4 5 6 7 8 9 10 11

82 78 87 89 81 74 80 87 91 62 100

Thoracic aortic aneurysm Myocardial infarction Cerebrovascular accidient Acute cardiac failure/renal failure Intra cerebral hemorrhage Myocardial infarction Bilateral cerebrovascular disease Congestive cardiac failure Acute pulmonary edema Severe pulmonary atherosclerosis Acute left ventricular failure

the bone-specific protein, osteopontin, by osteocytes as a trigger of remodeling induced by mechanical stress.24 Previously, we have shown that femoral neck osteocyte lacunar occupancy (calculated as the ratio of identifiable cells to identifiable osteocyte lacunae) was not significantly different between controls and patients in their seventh or eighth decade who had suffered an osteoporotic hip fracture. A positive association was observed, however, between osteocyte viability and regions of relatively high cortical remodeling.23 The aim of the present investigation was first to determine whether the hypothesis that osteocytes disappear from their lacunae with advancing tissue age could be substantiated in osteonal bone. To do this, the lacunar occupancy and osteocyte density within quiescent osteons were compared with that of forming osteons. Second, these parameters were compared in bone surrounding quiescent and resorbing osteons so as to test the Martin hypothesis.17 The statistical influence of the distances from the canal surface on lacunar density, osteocyte density, and lacunar occupancy was also investigated. Materials and Methods Subjects Samples used in the study were obtained from 22 subjects. Eleven were female patients who suffered intracapsular fracture of the femoral neck (mean age: 82.2 ⫾ 3.0 years). The biopsies were taken 1– 4 days after fracture. Informed written, witnessed consent, as approved by the local ethics committee, was obtained from each subject. Nonfractured, age-matched control material was obtained from 11 female postmortem cases (mean age: 82.2 ⫾ 1.4) (Table 1). The postmortems were held between 1 and 4 days of death and the tissue was processed within 3 h of postmortem. So as to avoid the early effects of disuse osteoporosis, individuals were not included if they had been admitted to hospital for ⬎14 days prior to death or if they had been admitted from other hospitals or residential care. In addition, subjects were screened to exclude from the study cases with a history of diseases such as carcinoma, which might affect femoral neck bone locally, or who had been prescribed drugs known to affect bone metabolism. Histology Femoral neck case and control samples were fixed in 4% paraformaldehyde for 12 h, dehydrated in ethanol, and cleared in toluene for embedding in methylmethacrylate. Sections were cut at 10 ␮m on a Jung Polycut E microtome (Leica, Milton Keynes, UK) and floated onto distilled water prior to

Figure 1. Phase-contrast micrographs of Goldner-stained tissue with superimposed grided template. Scale bar ⫽ 65 ␮m. (a) Quiescent haversian canal. (b) Forming haversian canal. (c) Resorbing haversian canal.

staining by a modified Goldner method. In the first instance, for a group of six fracture and six control samples, a total of 18 microscopic fields representing six resorbing (exhibiting a crenellated surface), six forming (osteoid bearing), and six quiescent haversian canals (Figure 1a– c) were identified in sequence and selected for capture using a low-light camera (Model LC-100C, Seescan). An additional five fracture and five control subjects were also included in the study where only six resorbing and six quiescent canals were captured per biopsy. Each field (area 0.18 mm2) was captured sequentially under phase-contrast, bright-field, and epifluorescence microscopy as previously described.23

Bone Vol. 30, No. 6 June 2002:859 – 865

J. Power et al. Osteocytes in quiescent and remodeling osteons

861

Table 2. Mean ⫾ SEM data for the parameters lacunar density (Lac/mm2), osteocyte density (Ost/mm2), and percentage lacunar occupancy (% occ) Cases of hip fracture Canal type a

Forming Resorbing Quiescent a

n 6 11 11

2

Lac/mm

932 ⫾ 54 806 ⫾ 40 750 ⫾ 33

2

Ost/mm

647 ⫾ 34 559 ⫾ 28 470 ⫾ 22

Controls % occ 70.6 ⫾ 2.8 69.7 ⫾ 2.0 63.6 ⫾ 3.3

n 6 11 11

2

Ost/mm2

c

b

Lac/mm

795 ⫾ 35 719 ⫾ 37 680 ⫾ 43

538 ⫾ 24b 463 ⫾ 31 401 ⫾ 34

% occ 68.5 ⫾ 3.4 64.6 ⫾ 3.2 59.3 ⫾ 3.1

Data from rings 2–5. Values are significantly lower than fracture group (p ⬍ 0.05) by analysis of variance (ANOVA). p ⫽ 0.06.

b c

Image Analysis Images were transferred to a Power Macintosh 9600/200 and scale calibrated (1537.5 pixels/mm) for analysis by NIH IMAGE (v1.61) software (developed at the U.S. National Institutes of Health and available on the internet by anonymous ftp from zippy.nimh.gov or on floppy disk from the National Information Service, Springfield, VA [part number PB95-500195GEI]). Individual gray-level fields were converted to binary images and the Haversian systems were expanded a pre-set distance (20 pixels, 13.0 ␮m) for five iterations to generate a template, centered at the canal surface. This distance was a value generated by the NIH IMAGE software and considered adequate in size, in the majority of cases, for the fitting of osteocyte lacunae, parallel to their long axis, within a width of 13 ␮m. Furthermore, in a previous investigation (based on a different subset of our current library of fracture [n ⫽ 12] and control biopsies [n ⫽ 11]), wall widths from cortical osteons were measured using polarized light.2 The mean value for wall width from a total of 1457 canals was found to be 42.8 ⫾ 0.46 ␮m (median 40.3 ␮m). Eighty-eight percent of the wall width distribution fell within a distance of 65 ␮m from the canal surface, so this was selected in the current study as the cut-off value for the extent of osteonal bone. Templates were then superimposed onto phase and ultraviolet (UV) images to quantify osteocyte lacunar number and osteocyte cell number present within each of the five concentric rings surrounding the osteon of interest (Figure 1a– c). The proportion of osteocytes remaining in their lacunae (percentage lacunar occupancy) was estimated from the quotient of the number of osteocyte nuclei visualized under UV within a given ring divided by the number of lacunae visualized under phase contrast. Areas of overlap caused by neighboring haversian canals on the central osteon of interest were discarded from the analysis. Scale calibration and measurement of bone areas within each individual ring allowed for the calculation of lacunar and osteocyte densities. Statistical Analysis The data were statistically analyzed using commercially available software (JMP statistical package, v3.1.6, SAS Institute, Cary, NC). Initially, data were available from 12 subjects (6 fracture cases and 6 controls) where, for each subject, images of 18 osteons were analyzed (6 resorbing, 6 forming, and 6 quiescent canals). Subdivision of osteonal bone surrounding each individual canal allowed for the generation of data for 90 “osteonal rings” per subject (5 rings per osteon). In addition, a further 10 subjects were included (5 fracture cases and 5 controls) where 12 osteonal images per subject were analyzed (6 resorbing and 6 quiescent canals) providing an additional 60 rings per subject. Overall mean values of lacunar density, osteocyte density, and percentage lacunar occupancy (⫾SEM) were calculated from each subject and represented the values for bone

extending from the canal surface of each of the three canal types to a maximal radius of 65 ␮m. Case-control differences relating to these parameters were tested using analysis of variance (ANOVA). To examine further case-control effects between the parameters of lacunar density, osteocyte density, and lacunar occupancy, data relating to quiescent osteons were subtracted from the values from forming and resorbing osteons, respectively. These data, representing the differences between remodeling osteons (forming and resorbing) and resting osteons, were then analyzed by ANOVA to compare fracture cases and controls. Differences in lacunar density, osteocyte density, and lacunar occupancy between forming and quiescent osteons were tested within the osteonal bone surrounding the mean of six resting and six forming canals from each of the 17 subjects (6 forming, 11 for quiescent) by single ANOVA. Ring 1 (the area adjacent to the canal surface) was excluded from the analysis due to impairment of osteocyte detection caused by the background fluorescence generated by osteoid. The combined values for rings 2–5 were calculated and the data again meaned to provide a single value per subject. Differences in lacunar density, osteocyte density, and lacunar occupancy between resorbing and quiescent osteons were assessed where the data comprising rings 1–5 were combined then meaned to generate a single value for each subject (n ⫽ 22). Significance was tested using paired t-tests. Assessment of the effect of osteocyte density lacunar density and lacunar occupancy with increasing distance from the canal surface was conducted as follows: Cell distributions over the five rings in quiescent and resorbing osteons were compared, and the data for each ring (1–5) were analyzed separately and meaned according to osteon category (resorbing or quiescent) to provide a single value for each ring per subject. A similar analysis comparing forming with quiescent osteons was performed on data from rings 2–5. Ring 1 was excluded due to the impairment of osteocyte detection by osteoid. Significant differences in the parameters of lacunar density, osteocyte density, and lacunar occupancy between quiescent and resorbing osteons were determined by paired t-test. To examine the possible influence of canal size on osteocyte density, lacunar density, and lacunar occupancy, the data sets representing the mean values for each of the five rings were divided into quartiles based on the overall distribution of canal areas. These parameters were each analyzed separately as variables dependent on canal size quartile by least squares regression analysis in models that included disease, canal type, and intersubject variation as independent covariates. Results Case-Control Differences The data presented in Table 2 represent the overall mean values (mean of six canals, generating a single value for each osteon

862

J. Power et al. Osteocytes in quiescent and remodeling osteons

Bone Vol. 30, No. 6 June 2002:859 – 865

category [i.e., resorbing, forming, and quiescent osteons]) of lacunar density, osteocyte density, and percentage lacunar occupancy (⫾SEM). Of the canal types analyzed by single ANOVA only in the forming canals did lacunar density approach a significantly higher value in the fracture group compared with controls (p ⫽ 0.06). With regard to osteocyte density, both forming (p ⫽ 0.024) and resorbing (p ⫽ 0.034) canals showed significantly lower values in the control group compared with the fracture group. There were no statistically significant effects of case vs. control for any of the three canal types when percent lacunar occupancy was analyzed. When the mean values for quiescent osteons were subtracted from those for forming and resorbing osteons, respectively, and the differences compared by ANOVA, no significant differences were detected for lacunar density (forming vs. quiescent, p ⫽ 0.7; resorbing vs. quiescent, p ⫽ 0.77), osteocyte density (forming vs. quiescent, p ⫽ 0.33; resorbing vs. quiescent, p ⫽ 0.56), and lacunar occupancy (forming vs. quiescent, p ⫽ 0.35; resorbing vs. quiescent, p ⫽ 0.87) between fracture cases and controls. Based on these observations, data from both fracture and control categories were combined in appropriate analyses of variance. Effect of Relative Tissue Age on Osteocyte Lacunar Occupancy, Osteocyte Density, and Lacunar Density (Comparison of Forming Vs. Quiescent Osteons) Lacunar occupancy values showed a significantly greater loss of osteocytes from their lacunae in resting (quiescent) osteons compared with the more recently deposited bone surrounding actively forming osteons (mean difference ⫾SEM: 8.1 ⫾ 3.4%; p ⫽ 0.025). Osteocyte density surrounding forming canals was also significantly greater than that of quiescent osteons (mean difference: 157 ⫾ 34/mm2; p ⬍ 0.0001). Lacunar density was significantly higher in the forming compared with quiescent osteons (mean difference: 148 ⫾ 46/mm2; p ⫽ 0.0031). Differences in Osteocyte Density, Lacunar Density and Percent Lacunar Occupancy Between Resorbing and Quiescent Osteons Resorbing osteons showed significantly higher osteocyte densities than quiescent osteons (mean difference: 76 ⫾ 23/mm2; p ⫽ 0.003). Lacunar density was found to be marginally higher in resorbing compared with quiescent osteons (mean difference: 47 ⫾ 26/mm2; p ⫽ 0.081). Resorbing canals demonstrated a significantly higher percent lacunar occupancy than quiescent canals (mean difference: 5.7 ⫾ 2.3%; p ⫽ 0.022). Effect of Distance From Canal Surface No statistical differences were observed in lacunar density in quiescent vs. resorbing osteons in any of the five rings (Figure 2a), nor were significant differences observed in osteocyte density in the first two rings (0 –13; 14 –26 ␮m from canal surface) between resorbing and quiescent osteons. In each of the three further peripheral rings, however, osteocyte density was significantly higher in resorbing compared with quiescent canals (Figure 2b; Table 3). For percent lacunar occupancy, when data from both cases and controls were combined, the only significant difference between resorbing and quiescent canals was in ring 5 (53– 65 ␮m from the canal surface) where occupancy was elevated (mean difference: ⫹8.8 ⫾ 3.2%; p ⫽ 0.013). Forming and quiescent osteons (rings 2–5) are compared in Figure 3 and no significant differences were seen for lacunar density in rings 2– 4. In ring 5, however, lacunar density was

Figure 2. Effect of distance from the canal surface: Comparison of quiescent (lightly stippled bars) and resorbing (densely stippled bars) osteons. *p ⬍ 0.05; **p ⬍ 0.01. (a) Lacunar density. (b) Osteocyte density. (c) Percent lacunar occupancy.

significantly higher in forming osteons (Figure 3a). Osteocyte density was significantly greater in forming vs. quiescent canals in each of rings 2–5 (Figure 3b). With regard to lacunar occupancy, only in ring 3 (40 –52 ␮m) were the levels not significantly elevated in forming osteons (Figure 3c). Statistical Effect of Canal Area Osteocyte and lacunar densities were quantified radially from a soft:hard tissue boundary defined by the canal surface. No

Bone Vol. 30, No. 6 June 2002:859 – 865

J. Power et al. Osteocytes in quiescent and remodeling osteons

863

Table 4. Mean ⫾ SEM data for lacunar density (Lac/mm2), osteocyte density (Ost/mm2), and percentage lacunar occupancy (% occ) for canal size quartiles 1– 4 Quartile

Lac/mm2

Ost/mm2

% occ

1 2 3 4

813 ⫾ 26 765 ⫾ 20 707 ⫾ 21 733 ⫾ 18

491 ⫾ 21 517 ⫾ 18 468 ⫾ 17 497 ⫾ 17

60.6 ⫾ 1.9 68.0 ⫾ 1.8 66.5 ⫾ 1.6 67.8 ⫾ 1.5

large Haversian canals, the data would be influenced by location of cells within tissue deeper relative to the canal center than in the case of smaller canals. To assess this potential influence, the median canal areas for the whole data set were divided into quartiles (median 8801 ␮m2, interquartile range 3358 –17,456 ␮m2; canal-type medians: quiescent 4685 ␮m2, forming 5364 ␮m2, resorbing 15,040 ␮m2). When canal size, categorized by size quartile, was entered as a covariate in the models it had no significant effect on the models for osteocyte density (p ⫽ 0.57) and lacunar occupancy (p ⫽ 0.25). There was, however, a marginally significant negative effect of canal size quartile on lacunar density (p ⫽ 0.054). Mean values of lacunar density, osteocyte density, and lacunar occupancy for canal size quartiles 1– 4 are shown in Table 4. Discussion

Figure 3. Effect of distance from the canal surface: Comparison of quiescent (lightly stippled bars) and forming (densely stippled bars) osteons. *p ⬍ 0.05; **p ⬍ 0.01. (a) Lacunar density. (b) Osteocyte density. (c) Percent lacunar occupancy.

assessment could be made of the position of the cement line in this study because it was not feasible to use polarized light for its identification. Therefore, it appeared possible that, in the case of

Previously, we reported a positive association between lacunar occupancy and regional indices of bone turnover within randomly selected cortical fields, including both osteonal and interstitial bone.23 To examine this association in more detail we analyzed the local osteocyte populations within the osteonal bone adjacent to both remodeling (forming and resorbing) and resting (quiescent) Haversian canals. For the first time, an assessment of the effect of distance from the Haversian canal surface upon the viability of osteocytes was also undertaken. There were no significant differences in lacunar occupancy between fracture and control subjects, suggesting that osteocyte viability rates were similar for osteonal bone in these groups. The overall mean lacunar occupancy value of 65.7% (⫾1.3%) observed in the present study differs from that previously reported by us (51.4 ⫾ 2.7%).23 This disparity is almost certainly a consequence of the almost exclusive sampling of osteonal bone compared with the more random approach adopted previously where interstitial bone, which is known to have lower lacunar occupancy, would have made a stronger contribution.26 Osteocyte density from fracture subjects within bone in forming and resorbing osteons and lacunar density in the forming osteons were significantly elevated compared with that of the control group. This increased osteocyte density found within osteonal bone of fracture subjects may reflect a higher rate of osteoblast incorporation into the matrix or a reduced rate of matrix apposition, for which there is evidence.1 Either way, the active bone-forming lifespan of the individual osteoblasts, prior

Table 3. Osteocyte density rings 3–5 showing p values for resorbing vs. quiesent canals p (mean difference ⫾ SEM/mm2)

Case ⫹ control Cases Control

Ring 3 (27–39 ␮m)

Ring 4 (40–52 ␮m)

Ring 5 (53–65 ␮m)

0.032 (⫹89 ⫾ 39) 0.025 (⫹119 ⫾ 45) 0.37 (⫹60 ⫾ 65)

0.019 (⫹77 ⫾ 30) 0.17 (⫹64 ⫾ 43) 0.068 (⫹91 ⫾ 44)

0.0026 (⫹115 ⫾ 34) 0.042 (⫹125 ⫾ 54) 0.035 (⫹105 ⫾ 43)

864

J. Power et al. Osteocytes in quiescent and remodeling osteons

to their incorporation as osteocytes, might be depressed in the fracture subjects compared with the control group. Consistent with this hypothesis, a previous study on osteocyte density in the cancellous bone of the ilium has shown elevated osteocyte densities in cases of osteoporosis.19 Alternatively, secondary hyperparathyroidism due to hypovitaminosis D, which results in a reduction in matrix mineralization by osteoblasts, has been linked with hip fracture,5 so impaired matrix mineralization might also lead to an elevated osteocyte density. Unfortunately, we did not have 25-hydroxyvitamin D plasma levels available for both controls and cases to test this hypothesis. A further finding was that lacunar densities, in both cases and controls, were higher in the forming than in the other two osteon categories, both of which comprised substantially older bone. If, as hypothesized separately by Martin,17 Noble et al.,20 and others,3,11,27 osteocytes are important in the regulation of bone remodeling, then their matrix density (assessed here as numbers per unit area of osteonal bone) might be a key determinant of this process. Osteocyte-rich tissue associated with actively forming canals as seen here, was an anticipated observation, reflecting the youth of these new packets of bone, and is consistent with Parfitt’s aging theory of osteocyte death.22 The observation that bone possessing relatively high osteocyte densities is also associated with resorption is a new and important finding. A recent theory proposed by Martin17 suggested that basal inhibition by osteocytes of the initiation of bone remodeling is a major unidirectional function of live osteocytes. Our observations suggest that this hypothesis requires reexamination to accommodate the possibility that osteocytes might promote as well as inhibit osteoclastic bone resorption, perhaps through multiple mechanisms.21 The distance of a given cell from sources of nutrients is thought to have an important influence on its ability to survive. In the case of the bone cell population of the cortex, the Haversian canal surface and its associated blood supply represents its primary potential nutrient source. As predicted from the concept that canalicular flows could be rate-limiting,13,14 the levels of osteocyte density reported in this investigation were seen to decline with distance from the quiescent canal surface. Osteocytes have been shown to be capable of expressing HIF-1, a marker of hypoxia.9 In resorbing osteons the density of these cells was, however, maintained with increasing distance from the canal surface (rings 3–5), more successfully than in quiescent osteons. Lacunar occupancy was significantly elevated in resorbing vs. quiescent canals in only the outermost ring. It is possible that the maintenance of viable osteocytes and their communications through the canalicular system might be necessary for the signaling required to initiate (or to limit) osteoclastic activity. Unsurprisingly, levels of osteocyte density and lacunar occupancy were generally higher in forming vs. quiescent systems. These findings are likely a reflection of the relative youth of the bone packets associated with forming systems. This study has a number of possible limitations. The numbers of cases and controls available for study were limited, affecting the power of our study to determine a significant difference between groups. For example, at a mean between-group difference of 70 lacunae/mm2, 58 cases and controls would have been needed to find a significant difference in lacunar density in resting osteons. Next, only a two-dimensional quantitation of the local osteocyte population associated with resting and remodeling osteons was undertaken. To avoid artifacts associated with differences in mean size in lacunar volume between groups, it would be necessary to use the disector principal applied to adjacent sections10 or confocal microscopy. We believe it to be unlikely, however, that there were major differences in lacunar or osteocyte cellular volumes sufficient to have a major influence

Bone Vol. 30, No. 6 June 2002:859 – 865

on our results. We tried to avoid unconscious selection bias by selecting fields free of artifacts, which were flat (so that all structures analyzed were as close as possible to one focal plane) and by using a phase contrast of ⫻10 objective magnification for the selection process. This made it impossible to identify lacunar contents prior to image capture. The generation of artifacts during processing, such as by dragging the contents out of some lacunae during sectioning, may have also influenced values of lacunar occupancy and osteocyte density; however, there is no reason to believe this would have affected one group more than another. Because the use of tetracycline labels was not feasible in this study, no discrimination could be made between osteoid undergoing active mineralization and resting osteoid. The presence of a crenellated osteonal surface was applied as the defining criterion for active resorption in this study. This method of classification could represent a further limitation because it is possible that a proportion of these canals had ceased osteoclastic resorption. On the other hand, the basic multicellular unit (BMU) theory of bone remodeling, first described by Frost,8 indicates that, if not actively resorbing, such crenellated canals would have been resorbing in the recent past in most cases and would exist in a state of preparation for bone formation. Enzyme histochemistry, such as staining for tartrate-resistant acid phosphatase to visualize sites of resorption used in conjunction with in vivo assessment of osteocyte viability, might provide further evidence of the relationship of osteocytes to biochemical expression of the remodeling process. Further studies involving immunohistochemistry are required to assess the expression of candidate molecules, such as osteopontin,24 which are implicated in the osteocytic regulation of bone remodeling. In conclusion, this investigation is the first attempt to quantify, at the level of the BMU, the local osteocyte distributions associated with both remodeling and resting Haversian canals. These data demonstrate that, whether an osteon is undergoing current resorption or has undergone recent resorption, it is statistically associated with higher osteocyte density than quiescent osteons. This finding challenges the hypothesis that attributes to osteocytes a regulatory role limited to the inhibition of bone remodeling. Our data encourage further investigation of key mechanisms whereby osteocytes might sometimes promote, and other times suppress, the resorption and remodeling of compact bone.

Acknowledgments: This work was supported by an MRC Programme Grant (9321536) and was presented in part at the 2000 Meeting of the Bone and Tooth Society.

References 1. Arlot M., Edouard C., Meunier P. J., Neer R. M., and Reeve J. Impaired osteoblast function in osteoporosis. Comparison between calcium balance and dynamic histomorphometry. BMJ 289:517–520; 1984. 2. Bell K. L., Loveridge N., Jordan G. R., Power J., Constant C. R., and Reeve J. A novel mechanism for the induction of the increased cortical porosity in cases of intracapsular hip fracture. Bone 27:297–304; 2000. 3. Burr D., Forwood M., Fyhrie D., Martin R., Schaffler M., and Turner C. Bone microdamage and skeletal fragility in osteoporotic and stress fractures. J Bone Miner Res 12:6 –15; 1997. 4. Carter D. R. Mechanical loading history and skeletal biology. J Biomech 20:1095–1109; 1987. 5. Chapuy M. C., Arlot M. E., Delmas P. D., and Meunier P. J. Effect of calcium and cholecalciferol treatment for three years on hip fractures in elderly women. Br Med J 308:1081–1082; 1994. 6. Compston J. E., Yamaguchi K., Croucher P. I., Garrahan N. J., Lindsay P. C., and Shaw R. W. The effects of gonadotrophin-releasing hormone agonists on

Bone Vol. 30, No. 6 June 2002:859 – 865

7.

8. 9.

10.

11.

12.

13.

14.

15. 16.

17. 18. 19.

iliac crest cancellous bone structure in women with endometriosis. Bone 16:261–268; 1995. Duncan R. L., and Misler S. Voltage-activated and stretch-activated Ba2⫹ conducting channels in an osteoblast-like cell line (UMR-106). Fed Eur Bone Soc Lett 251:17–21; 1989. Frost H. M. Tetracycline-base histological analysis of bone remodeling. Calcif Tissue Res 3:211–237; 1969. Gross T. S., Akeno N. A., Clemens T. L., Komarova S., Srinivasan S., Weimer D. A., and Mayorov S. Physiological and genomic consequences of intermittent hypoxia. Selected contribution: Osteocytes upregulate HIF-1 alpha in response to acute disuse and oxygen deprivation. J Appl Physiol 90:2514 –2519; 2001. Gundersen H. J. Stereology of arbitory particles. A review of unbiased number and size estimators and the presentation of some new ones, in memory of William R Thompson. J Microsc 143:3–45; 1986. Huiskes R., Wienans H., Grootenboer H. J., Dalstra M., Fudala B., and Sloof T. J. Adaptive bone remodeling theory applied to prosthetic-design analysis. J Biomech 20:1135–1150; 1987. Klein-Nulend J., Van der Plas A., Semeins C. M., Ajubi N. E., Frangos J. A., Nijweide P. J., and Burger E. H. Sensitivity of osteocytes to biomechanical stress in vivo. FASEB J 9:441–445; 1995. Knothe Tate M. L., Niederer P., and Knothe U. In vivo tracer transport through the lacunocanalicular system of rat bone in an environment devoid of mechanical loading. Bone 22:107–117; 1998. Lozupone E., Nico B., Mancini L., Favia A., Lamanna M., and Cagiano R. The dynamic of the flow of the interstitial fluid into the osteocyte lacunae and canaliculi of bone cultured in vitro. Bone 19(Suppl. 3):150S; 1996. Marotti G. The structure of bone tissues and the cellular control of their deposition. Ital J Anat Embryol 101:25–79; 1996. Marotti G., Ferretti M., Muglia M. A., Palumbo C., and Palazzini S. A. A quantitive evaluation of osteoblast-osteocyte relationships on growing endosteal surface of rabbit tibia. Bone 13:363–368; 1990. Martin R. B. Towards a unifying theory of bone remodeling. Bone 26:1–6; 2000. Mullender M. G., and Huiskes R. A. A proposal for the regulatory mechanism of Wolff’s law. J Orthopaed Res 13:503–512; 1995. Mullender M. G., Meer D. D., Huiskes R., and Lips P. Osteocyte density changes in ageing and osteoporosis. Bone 18:109 –113; 1996.

J. Power et al. Osteocytes in quiescent and remodeling osteons

865

20. Noble B., Stevens H., Loveridge N., and Reeve J. Identification of apoptotic changes in osteocytes in normal and pathological human bone. Bone 20:273– 282; 1997. 21. Noble B. S., and Reeve J. Osteocyte function, osteocyte death and bone fracture resistance. Mol Cell Endocrinol 159:7–13; 2000. 22. Parfitt A. Skeletal heterogeneity and the purposes of bone remodelling. Implications for the understanding of osteoporosis. In: Marcus R., Feldman D., & Kelsey J., eds. Osteoporosis. New York: Academic; 315–329; 1996. 23. Power J., Noble B. S., Loveridge N., Bell K. L., Rushton N., and Reeve J. Osteocyte lacunar occupancy in the femoral neck cortex: An association with cortical remodeling in hip fracture cases and controls. Calcif Tissue Int 69:13–19; 2001. 24. Terai K., Takano-Yamamoto T., Ohba Y., Hiura K., Sugimoto M., Sato M., Kawahata H., Inaguma N., Kitamura Y., and Nomura S. Role of osteopontin in bone remodeling caused by mechanical stress. J Bone Miner Res 14:839 –849; 1999. 25. Tomkinson A., Reeve J., Shaw R. W., and Noble B. S. The death of osteocytes via apoptosis accompanies estrogen withdrawal in human bone. J Clin Endocrinol Metab 82:3128 –3135; 1997. 26. Urist M. R., MacDonald N., Moss M., and Skoog W. Rarefying disease of the skeleton: Observations dealing with aged and dead bone in patients with osteoporosis. In: Sognnaes R. F., ed. Mechanisms of Hard Tissue Destruction. Washington, DC: American Association for the Advancement of Science; 385–446; 1963. 27. Verborgt O., Gibson G. J., and Schaffler M. B. Loss of osteocyte integrity in association with microdamage and bone remodeling after fatigue in vivo. J Bone Miner Res 15:60 –67; 2000. 28. Weinbaum S., Cowin S. C., and Zeng Y. A model for the excitation of osteocytes by mechanical loading-induced bone fluid shear stresses. J Biomech 27:339 –360; 1994.

Date Received: January 23, 2001 Date Revised: December 11, 2001 Date Accepted: January 23, 2002