Presented in Oleksik A, Ott S, Vedi S, Compston J, Lips P 1998 Bone structure in patients with low bone mineral density with or without vertebral fractures (abstract). Bone 23:S520(F464).
Bone Structure in Patients with Low Bone Mineral Density With or Without Vertebral Fractures†
Article first published online: 1 JUL 2000
Copyright © 2000 ASBMR
Journal of Bone and Mineral Research
Volume 15, Issue 7, pages 1368–1375, July 2000
How to Cite
Oleksik, A., Ott, S. M., Vedi, S., Bravenboer, N., Compston, J. and Lips, P. (2000), Bone Structure in Patients with Low Bone Mineral Density With or Without Vertebral Fractures. J Bone Miner Res, 15: 1368–1375. doi: 10.1359/jbmr.2000.15.7.1368
- Issue published online: 2 DEC 2009
- Article first published online: 1 JUL 2000
- Manuscript Accepted: 28 JAN 2000
- Manuscript Revised: 9 DEC 1999
- Manuscript Received: 23 JUL 1999
- cancellous bone structure;
- strut analysis;
- cortical thickness;
- prevalent vertebral fractures;
- postmenopausal osteoporosis
Vertebral fractures (VFX) are caused by low bone mass and microstructural deterioration of bone tissue. The latter is not well defined. We investigated bone structure in transiliac biopsy specimens from 88 volunteers. Biopsy specimens were obtained at baseline in the Multiple Outcomes of Raloxifene Evaluation trial, a prospective study in osteoporotic (BMD ≤ −2.5 T score) postmenopausal women without or with VFX on standardized lateral spinal radiographs. Bone biopsy specimens were embedded in methylmethacrylate (MMA). Histomorphometry was done in 8 μm (U.S.A.) or 5 μm (Europe) Goldner stained sections. Vertebral fracture status (yes/no) was the outcome variable in logistic regression models adjusted for age and biopsy specimen origin (U.S.A. vs. Europe). Patients with and without VFX (26/62) were similar regarding age (69.2 ± 5.2 years vs. 67.3 ± 6.7 years), bone volume (BV/TV; 17.7 ± 4.7% vs. 19.0 ± 5.8%), and bone surface (BS/TV; 2.7 ± 0.6 mm2/mm3 vs. 2.8 ± 0.6 mm2/mm3). A lower cortical thickness (C.Th; 652 ± 267 μm vs. 822 ± 325 μm), total strut length (TSL; 826 ± 226 μm/mm2 vs. 922 ± 256 μm/mm2), node-to-loop (Nd-Lp) strut length (10.1 ± 10.3% vs. 15.0 ± 13.6%), together with a higher node-to-terminus (Nd-Tm) strut length (45.6 ± 9.7% vs. 39.1 ± 9.3%) were each associated with prevalent VFX (0.01 < p < 0.10). Differences in BV/TV did not explain these associations. In conclusion, cortical thinning and disruption of trabecular lattice are possible pathogenic mechanisms in patients with VFX.
OSTEOPOROSIS IS a systemic skeletal disease characterized by low bone mass and microstructural deterioration of bone tissue with a consequent increase in bone fragility.(1) According to the World Health Organization (WHO) criteria, osteoporosis is diagnosed when bone mineral density (BMD) is lower than 2.5 SD below the young normal mean.(2) This disease has a great impact on quality of life because it results in fragility fractures of hip, distal forearm, and vertebrae. Fortunately, not all patients with osteoporosis will suffer fractures. This may be because of differences in the degree or in the pattern of architectural deterioration of cortical and/or cancellous bone causing differences in strength. This hypothesis, although plausible, has not been proven. It is uncertain whether bone structure in patients with osteoporotic fractures is qualitatively different from that in patients with age-related bone loss not having suffered fractures.(3–5)
Most authors have reported a decrease in cancellous bone volume (BV/TV) and trabecular number (Tb.N) and a consequent increase of trabecular separation (Tb.Sp) with aging and in patients with vertebral fractures (VFX).(6–10) These findings led to the conclusion that the loss of entire elements because of trabecular perforation forms the main mechanism of structural changes in primary osteoporosis. In some studies, trabecular thinning, which could be an alternative way of bone loss, was reported.(3, 6, 7) Studies on trabecular connectivity confirmed disruption of bone structure in patients with osteoporotic fractures(11, 3) although the difference from control subjects disappeared after matching for cancellous BV/TV.(3, 9)
Some investigators have assumed that postmenopausal osteoporosis is characterized by accelerated loss of preferentially cancellous bone and relative preservation of cortical bone.(2–15) However, recent estimations suggest that the absolute amount of bone lost is greater in cortical bone(16) resulting in cortical thinning. Cortical thickness (C.Th) was lower in postmenopausal women with VFX than in controls with normal BMD.(12, 17) The age-related decline of C.Th seems less evident in iliac biopsy specimens(12) than in the lumbar vertebrae.(17, 18)
In most of the above-mentioned studies on bone structure,(3, 7, 9–12, 17) patients with osteoporotic fractures (mostly vertebral) were older than control subjects and no simultaneous adjustment was made for differences in BV/TV and age (duration of “exposure” to decreased bone mass). In the present study we investigated cancellous bone structure and C.Th as (age- and BV/TV-adjusted) predictors of prevalent VFX in postmenopausal women with low BMD.
MATERIALS AND METHODS
Baseline transiliac biopsy specimens were studied from 88 volunteers who participated in the Multiple Outcomes of Raloxifene Evaluation (MORE) trial, a multicenter study on the effects of the selective estrogen receptor modulator raloxifene in ambulatory postmenopausal women with osteoporosis (n = 7705, aged up to 80 years). The main inclusion criterion was low BMD of the femoral neck or spine (T score ≤ −2.5). Patients with metabolic bone disease or malignancy were excluded. The women did not have antiosteoporotic therapy with the exception of calcium and vitamin D or estrogen up to 6 months before inclusion.
Before inclusion, standardized spinal X-rays were sent to the central radiological facility in order to detect vertebral deformities in a consistent semiquantitative way. The radiographs from European sites were sent to Kiel in Germany (Dr. C.C. Glüer) and those from American sites to San Francisco, CA (Dr. H. Genant). VFX were defined as anterior, central, or posterior height loss in comparison with the adjacent vertebrae. A fracture was defined as at least 20% height loss.(19) Subsequently, in all patients with at least one vertebral fracture and in a sample of patients without VFX two independent readings of digitized images were performed, binary semiquantitative (BSQ) and quantitative morphometry (QM), with both baseline and follow-up displayed in order to control for false positives. A vertebra was defined fractured when two of the three readings of the baseline X-rays agreed.(20) The BMD data are the means from two baseline measurements.
Bone biopsy protocol
The bone biopsy protocol was performed in four a priori selected centers and it was approved by the local ethical review boards. All MORE participants in those centers were asked to volunteer and the volunteers gave informed consent. Fifty-five biopsy specimens came from two centers in the United States and 33 from two centers in Europe. Before obtaining the bone biopsy specimens, the patients were labeled with 4 × 250 mg tetracycline (tissue time marker) per day according to the 2–10–2 schedule. The transiliac bone biopsy specimens were obtained from the anterior iliac crest under local anesthesia using trephines of 7.5 mm (8 mm in the U.S.A.) inner diameter.
The biopsy specimen was fixed for 24 h in 4% phosphate-buffered formaldehyde and then transferred to 70% alcohol. After dehydration in graded alcohols, the bone specimens were embedded without prior decalcification in methylmethacrylate (MMA; BDH Chemicals LTD, Poole, UK) supplemented with 20% plastoid-N (Röhm und Haas Pharma GmbH, Darmstadt, Germany) and 0.13 g/ml perkadox (AKZO Chemicals, Deventer, The Netherlands).(21) Sections of 5 μm (8 μm in the U.S.A.) were prepared using a Jung K microtome. Sections were stained with Goldner's trichrome or left unstained for fluorescence microscopy. European sections were stained after removal of MMA. All measurements were performed without knowledge of BMD or fracture status.
Bone mass and bone remodeling indices
Histomorphometry of the European biopsy specimems was done using a microscope fitted with a drawing tube (Leitz, Wetzlar, Germany), digitizer, and software (Zeiss Kontron, Oberkochen, Germany). For most measurements, Osteoplan software was used except for the dynamic variables osteoid thickness and C.Th, which were measured with Videoplan. The sections through the center of the core were measured first and when necessary, more sections, 200 μm apart, were measured. For bone mass and bone remodeling indices, at least 20 mm2 of cancellous bone were measured under ×125 magnification. Osteoid surface (OS/BS) was measured when the osteoid seam was at least one lamella thick. Osteoid thickness (O.Th; x200 magnification) and interlabel distance was measured directly using the Videoplan software and Zeiss integration eyepiece type II (Zeiss, Oberkochen, Germany) for random sampling of measurement sites according to Kragstrup et al.(22, 23) Measurements were done at 100 intercepts in each biopsy. The histomorphometry of the US biopsies was done using a microscope fitted with a drawing tube and the Osteometrics software (Osteometrics, Atlanta, GA, U.S.A.). The measurements were performed under ×100 (bone mass), ×160 (mineralizing surface MS/BS), and ×250 (distances) magnifications. Interobserver variance was assessed on 10 biopsy specimens, five from the United States and five from Europe. The coefficients of variation were similar to those reported by Compston (BV/TV, 10.3%; OS/BS, 22.6%; MS/BS, 29.7%; eroded surface (ES)/BS, 77.0%; mineral apposition rate (MAR), 28.0%).(24)
C.Th and endocortical envelope width were assessed by one observer (AO). The measurements were done only when both cortices were intact (n = 71). The mean of four measurements per cortex or per envelope was calculated. The nomenclature used and calculations performed were in accordance with the ASBMR Nomenclature Committee.(25) The following variables were calculated: trabecular thickness (Tb.Th), Tb.N, trabecular separation (parallel plate model), osteoid volume (OSO.Th), and bone formation rate (BFR/BS).
Measurements of bone structure were performed by one observer (AO) on four (Europe) or two (U.S.A.) Goldner stained sections at ×9 magnification using an image analyzer (IBAS II; Kontron, Germany) and software described earlier by Garrahan and Compston.(11, 26, 27) The measurements were performed on 75 biopsy specimens with both endocortical boundaries present. The section boundary lines were drawn manually on the image but the transition between cortical and endosteal regions was delineated automatically by two stages of binary dilatation. The cancellous BV/TV was measured automatically by pixel counting on the binary image. Trabecular surface was assessed by direct pixel counting and subtracting those that entered the cortex or upper or lower boundaries. Subsequently, the binary image containing upper and lower boundaries and corticomedullary delineation was skeletonized until the edge pixels of the bone were eroded away to a point at which further erosion would cause the bone to disappear (Fig. 1). Struts, termini (ends of struts), and nodes (junctions of struts) were identified according to defined topology (Fig. 1). Termini and nodes were expressed as counts/millimeter squared of tissue area and the total length of each strut type was expressed as a percentage of the total strut length (TSL). All struts entering the upper or lower boundaries were excluded from the analysis. Additionally, marrow star volume and trabecular bone pattern factor (TBPF) also were calculated.(28–30)
Statistics were performed using Statistical Package for Social Science (SPSS) software. The semiautomatic measurements of BV/TV were compared with results of automatic measurements. First, 95% limits of agreement were calculated between logs of the data. Afterward, the 95% confidence interval (CI) for the proportional agreement was calculated as the antilogs of limits of agreement. The semiautomatically estimated BV/TV was likely to be between 94.8% and 99.3% of the value that was obtained by automatic measurements (BV/TV(osteoplan/osteomeasure) = 1.41 + 0.95BV/TV(IBAS), R2 = 0.93, and p < 0.001). The 95% CI for the proportional agreement between semiautomatically and automatically obtained BS/TV was narrower (91.2–92.4%; BS/TV(osteoplan/ostomeasure) = 0.23 + 0.82BS/TV(IBAS), R2 = 0.93, and p < 0.001). The automatic measurements of BV/TV were used in the further statistical analysis, as they were obtained by one observer.
The data are expressed as range and tertiles for the total distribution and as mean ± 1 SD for distribution in the groups of patients without and with VFX. All correlation coefficients were corrected for method (partial correlation). Differences between biopsy specimens from the United States and Europe were studied using t-test for independent observations. The associations between variables and prevalent VFX were assessed using logistic regression models. Vertebral fracture status (yes or no) was the outcome variable. All models were adjusted for age and method (U.S.A. vs. Europe). Initially, the shape of a possible association between independent variable and presence of VFX was controlled by dividing the independent variable into tertiles of distribution that were entered into the model as indicator variables. Colinearity (correlation of estimates > 0.70) was avoided. The test statistics (maximum likelihood ratio test for the entire variable) was considered significant at p ≤ 0.10. In this way selected possible associations between the measurements of bone structure and prevalent VFX were additionally adjusted for BV/TV. When the association remained significant (maximum likelihood ratio test for the entire variable; p ≤ 0.10), the identified variables were used in multiple logistic regression models (age-, BV/TV, and method-adjusted). The combination of variables in the models was based on partial correlation coefficients after correction for age, BV/TV, and method. The final models were defined using backward stepwise elimination (maximum likelihood ratio test for the entire variable; p < 0.10).
Baseline characteristics, BMD, and regular histomorphometric data are presented in Table 1. From the measured bone mass and bone structure indices, only C.Th showed a significant correlation with age (r = − 0.32; p = 0.007). As expected, bone structure indices were correlated with BV/TV (p < 0.01). Bone formation rate was positively associated with Tb.Th and negatively with Tb.N, whereas erosion surface correlated positively with Tb.Sp (p < 0.05).
Patients from the United States, in comparison with those from Europe, had lower body height (159.7 ± 5.9 cm vs. 162.5 ± 5.0 cm; p = 0.03) and lower BMD in the trochanter (0.562 ± 0.072 g/cm2 vs. 0.598 ± 0.085 g/cm2; p = 0.05). The American and European biopsy specimens differed with regard to a number of histomorphometric indices (p < 0.05): CTh (700 ± 262 μm vs. 890 ± 366 μm), erosion surface (1.8 ± 1.1% vs. 5.7 ± 2.8%), osteoid thickness (9.0 ± 2.1 μm vs. 7.0 ± 1.9 μm), osteoid volume (1.8 ± 1.2% vs. 1.1 ± 0.8%), and wall thickness (W.Th 44.3 ± 6.2 μm vs. 34.9 ± 4.7 μm). The prevalence of VFX was similar in both study sites. Patients with and without VFX did not differ with regard to the assessed demographic data, bone remodeling, and bone mass parameters, with exception of Tb.Sp (Table 1).
Bone structure measurements were performed by one observer and are presented in Table 2. C.Th was 20% lower in patients with VFX than in those without VFX. There were no significant associations between presence of VFX and Tb.Sp, BV/TV, BS/TV, Tb.N, and Tb.Th (Table 2). Concerning strut analysis, patients with a lower TSL per section area, a lower node-to-loop (Nd-Lp) strut length or a higher node-to-terminus (Nd-Tm) strut length were more likely to have prevalent VFX (Table 2). Nd/Tm ratio and TBPF also showed significant associations with prevalent VFX. However, those associations were “bell-shaped.” When compared with reference categories (highest tertile for Nd/Tm ratio but lowest tertile for TBPF), the odd's ratios (ORs) for middle tertile (OR and 95% CI, respectively, 5.8 [1.4; 23.6] and 8.9 [2.0; 39.0]) were more prominent than the ORs for the difference between outer tertiles (OR and 95% CI, respectively, 2.1 [0.5; 8.8] and 2.8 [0.6; 12.6]).
The predictive value of bone structure parameters for the presence of VFX according to logistic regression is shown in Table 3. Differences in BV/TV did not explain the described associations between bone structure measurements and prevalence of VFX. The measurements of trabecular structure were interrelated. As expected, there was a particularly strong association (r = −0.65; p < 0.001) between TBPF and Nd-Tm ratio. Nd-Tm ratio was associated with TSL (r = 0.27, p < 0.05) and strongly associated with Nd-Tm strut length, and Nd-Lp strut length (r = −0.51 and r = 0.54, respectively; p < 0.001). C.Th was not associated with the indices of cancellous bone structure.
Based on partial correlation coefficients, multiple logistic regression models were constructed, all containing C.Th and one of the indices of trabecular structure. After backward stepwise elimination of bone structure indices (p out ≥ 0.10), C.Th remained significant in all models (4.6 < OR < 5.7 and p < 0.03) together with one index of trabecular structure (Nd/Tm ratio, OR = 16.5 and p = 0.01; TSL, OR = 7.4 and p = 0.04; Nd-Tm, OR = 4.4 and p = 0.03; Nd-Lp, OR = 4.3 and p = 0.06; TBPF, OR = 8.4 and p = 0.01, respectively).
Bone structure in patients with VFX
We have found that disruption of the trabecular lattice is associated with VFX. Specifically, a lower TSL, a higher TBPF (a higher proportion of convex structures), and a lower Nd-Tm ratio were associated with VFX whereas Tb.Th was not. A higher Tb.Sp was also associated with VFX but, probably a result of smaller group size, the association lost its significance when the analysis was restricted to the data obtained by one observer.
In terms of strut analysis, a shift from Nd-Lp strut length into Nd-Tm strut length was a risk factor for VFX, suggesting a specific pattern in the loss of trabecular elements. Absence of association between node-to-node (Nd-Nd) strut length and VFX may be explained as follows: when loops become disrupted, the Nd-Lp struts are replaced by Nd-Nd struts; when the deterioration progresses, the Nd-Nd struts are replaced by Nd-Tm struts. Similarly, disruption of the trabecular lattice does not necessarily implicate an increase in the terminus-to-terminus (Tm-Tm) strut length, which is a two-dimensional representation of rods rather than plates. Change from plate to rod will increase while elimination of rods will decrease this index. Tm-Tm strut length is likely to be increased in osteoporotic patients when compared with younger controls.(3, 9) However, in one study, after matching by BV/TV, biopsy specimens from postmenopausal patients with vertebral fracture and control subjects had similar Tm-Tm strut length,(3) supporting our results.
In this study, a lower C.Th was strongly associated with VFX. This confirms the important role of the cortical thinning in the pathogenesis of vertebral fracture.(12, 16–18, 31) The contribution of cortical bone loss to the pathogenesis of VFX may be less than that of disruption of cancellous bone structure, but this is a controversial issue.(17, 32–35) In our models, cortical thinning and indices of bone structure are confounders, because they are related to the fracture outcome and to each other; cortical thinning is enhanced by confluence of subendocortical cavities and perforation to the trabecular space.(16)
Measurement of C.Th to predict VFX may be of practical clinical importance because C.Th can be assessed by noninvasive procedures such as quantitative computed tomography(36) and because loss of cortical bone in vertebrae is only weakly correlated with changes in BMD.(37) The role of C.Th for prediction of new VFX should be assessed in a prospective setting.
Limitations of this study
The relevance of transiliac biopsies for the study of VFX has been previously discussed.(16, 18, 38–48) First, the ratio between the volume of cortical and cancellous bone is much lower in the vertebra than in the iliac crest.(16) Moreover, when histomorphometric indices of the iliac crest and of the lumbar spine were compared, the iliac crest had similar(45) or higher cancellous BV/TV,(39) lower Tb.Sp,(39) higher node and less terminus counts,(43) and finally, a lower terminus-related strut length.(39) On the other hand, strong correlations between the vertebrae and iliac crest were observed for marrow star volume,(28) node count(43) and Nd-Nd strut length.(43) Finally, in vertebral as well as in iliac bone, the loss of entire elements has been proposed as a mechanism of age-related bone loss.
As expected, we found that TBPF and Nd-Tm ratio were associated with prevalence of VFX.(29) However, these associations were bell-shaped, probably because both indices do not take into account the density and the distribution (equal vs. locally concentrated) of the assessed elements within a biopsy specimen.(26, 29) For example, biopsy specimens with a high TBPF contain mostly nonconnected elements. Biopsy specimens with only few nonconnected elements have similar TBPF as those with many well distributed nonconnected elements. Moreover, the later type of biopsy specimen has a higher TBPF than biopsy specimens with only a few, locally concentrated but well-interconnected elements. In general, because the two-dimensional approach to cancellous bone connectivity does not depict the trabecular shape, localization, and direction, it does not necessarily correlate with the strength of bone.(26) However, connectivity variables measured in the second lumbar vertebra correlated well with ultimate compressive stress.(43)
Our study was limited to a well-defined population of postmenopausal women with low BMD. VFX were a selection criterion for one of the substudies but not for the biopsy protocol, allowing us to use the VFX status as a dependent variable. This adds to the previous studies in which bone structure indices were used as dependent variables.(3, 7, 9–12, 17) Because of similar demographic data in the groups of patients with and without VFX, this design is suited for a study on specific structural differences. However, the differences in bone structure are difficult to detect because of an overlap between patients with and without VFX; some patients without fractures may develop one in the near future. This, together with complexity of the models, could contribute to the wide CIs for the estimated ORs. The sensitivity of the analysis could be improved by adjusting for other relevant variables such as vertebral geometry, mean degree of mineralization, and degree of trauma.
All logistic regression models were adjusted for method (biopsy specimens from Europe vs. the U.S.A.) and eventually for interaction with method. This procedure was necessary because the bone tissue preparation, histomorphometry, and the reading of the spinal radiographs were performed in two different centers for American and European subjects.
Patients without and with VFX were not different with regard to bone remodeling, but the study was not designed for studying those differences. The patients with VFX may present a wide range of bone remodeling.(46) High bone remodeling may result in disruption of trabecular lattice and it decreases the degree of the mineralization and increases the proportion of low mineral osteons.(47, 48) Although low bone remodeling may be associated with an increased mineralization degree, it also may result in delayed repair of microfractures. Therefore, after restricting the study subjects to those with low BMD (as in this study), the expected differences in bone remodeling between patients without and with VFX are small and masked by the combined variance of the interobserver differences and differences in histological techniques. When the statistical analysis was performed separately in patients from Europe and those from United States, no trends toward association between bone remodeling and VFX were found (data not shown).
We conclude that in postmenopausal women with low BMD, prevalent VFX are associated with cortical thinning and disruption of the trabecular lattice. After adjustment for age, patients with VFX could be distinguished from those without by lower C.Th, TSL, Nd-Nd strut length, and Nd-Tm ratio together with higher Nd-Tm strut length, and TBPF. Differences in BV/TV did not explain the observed associations. This study confirms the importance of bone structure in addition to bone mass as a contributor to bone strength.
We are grateful to P. Holzman and M.A. Berrie for preparing histological sections; Dr. E. Eriksen, Dr. R.R. Recker, and Dr. R. Khairi for allowing us to study bone biopsy specimens from their patients; Dr. C.C. Glüer and Dr. H.K. Genant for reviewing the X-rays; and S. Pluijm and E. Tromp for statistical assistance.
- 1Anonymous 1993 Consensus development conference: Prophylaxis and treatment of osteoporosis. Am J Med 94:646–650.
- 2World Health Organization (WHO) Study Group 1994 Assessment of fracture risk and its application to screening for postmenopausal osteoporosis. WHO technical report series 843. Geneva, WHO.
- 31994 Structural mechanisms of trabecular bone loss in primary osteoporosis: Specific disease mechanism or early ageing? Bone Miner 25:111–121., ,
- 41988 Age-related changes in cancellous bone structure: A two-dimensional study in the transiliac and iliac crest biopsy sites. Bone Miner 4:197–216., , , , , ,
- 51989 Age-related changes in trabecular width and spacing in human iliac crest biopsies. Bone Miner 6:331–338., ,
- 61983 Relationships between surface, volume, and thickness of iliac trabecular bone in aging and in osteoporosis. J Clin Invest 72:1396–1409., , , , ,
- 71990 A comparison of iliac bone histomorphometric data in postmenopausal osteoporotic and normal subjects. Bone Miner 11:217–235., , , ,
- 81985 The role of three-dimensional trabecular microstructure in the pathogenesis of vertebral compression fractures. Calcif Tissue Int 37:594–597., , , ,
- 91992 Loss of trabecular connectivity in osteoporosis demonstrated with independent methods. Bone 13:A28., ,
- 101992 Structural variables in postmenopausal women: Comparison among normal women and patients with osteoporosis or primary hyperparathyroidism. Bone 13:A26., , , , , , , , , , ,
- 111986 A new method for the two-dimensional analysis of bone structure in human iliac crest biopsies. J Microsc 142:341–349., ,
- 121991 Structural and geometric changes in iliac bone: Relationship to normal aging and osteoporosis. J Bone Miner Res 6:759–766., , , ,
- 131998 A unitary model for involutional osteoporosis: Estrogen deficiency causes both type I and type II osteoporosis in postmenopausal women and contributes to bone loss in aging men. J Bone Miner Res 13:763–773., ,
- 141986 Involutional osteoporosis. N Engl J Med 314:1676–1686.,
- 151981 Differential changes in bone mineral density of the appendicular and axial skeleton with aging. J Clin Invest 67:328–335., , , , ,
- 161998 A structural approach to renal bone disease. J Bone Miner Res 13:1213–1220.
- 171997 The thickness of human vertebral cortical bone and its changes in aging and osteoporosis: A histomorphometric analysis of the complete spinal column from thirty-seven autopsy specimens. J Bone Miner Res 12:89–95., , , ,
- 181991 Biologically meaningful determinants of the in vitro strength of lumbar vertebrae. Bone 12:219–224., , , , , ,
- 191993 Vertebral fracture assessment using a semi-quantitative technique. J Bone Miner Res 8:1137–1148., , ,
- 201999 Reduction of vertebral fracture risk in postmenopausal women with osteoporosis treated with raloxifene: results from a 3-year randomized clinical trial. Multiple Outcomes of Raloxifene Evaluation (MORE) Investigators [see comments]. JAMA 282:637–645., , , , , , , , , , , , , , , , ,
- 211983 An improved method for embedding hard tissue in polymethyl methacrylate. Stain Technology 58:135–141.,
- 221982 Estimation of the three-dimensional wall thickness of completed remodeling sites in iliac trabecular bone. Metab Bone Dis Rel Res 4:113–119., , ,
- 231987 An analysis of bone structure in patients with hip fracture. Bone Miner 3:63–73., ,
- 241986 Inter-observer and intra-observer variation in bone histomorphometry. Calcif Tissue Int 38:67–70., ,
- 251987 Bone histomorphometry: Standardisation of nomenclature, symbols, and units. Report of the ASBMR histomorphometry nomenclature committee. J Bone Miner Res 2:595–610., , , , , , ,
- 261994 Connectivity of cancellous bone: Assessment and mechanical implications. Bone 15:463–466.
- 271993 Quantitative analysis of trabecular bone structure. Bone 14:187–192., , , ,
- 281990 Star volume of marrow space and trabeculae in iliac crest: Sampling procedure and correlation to star volume of first lumbar vertebra. Bone 11:149–155.
- 291992 Trabecular bone pattern factor—a new parameter for simple quantification of bone microarchitecture. Bone 13:327–330., , ,
- 301996 Assessment of cancellous bone structure: comparison of strut analysis, trabecular bone pattern factor, and marrow space star volume. J Bone Miner Res 11:955–961., ,
- 311990 Fracture risk: A role for compact bone. Calcif Tissue Int 47:191–193.
- 321988 Functional biomechanics of the thoracolumbar vertebral cortex. Clin Biomech 3:11–18., , , ,
- 331986 Normal vertebral bone size and compressive strength: Relations to age and to vertebral and iliac trabecular bone compressive strength. Bone 7:207–212.,
- 341969 The relative contribution of trabecular and cortical bone to the strength of human lumbar spine. Calcif Tissue Res 3:163–175., ,
- 351985 Prediction of vertebral compressive fracture using quantitative computed tomography. J Bone Joint Surg 67A:1206–1214., , , ,
- 361989 Reference values for trabecular and cortical vertebral bone density in single and dual-energy quantitative computed tomography. Eur J Radiol 9:75–80., , , , , ,
- 371988 Comparison of bone mass measured by histomorphometry on iliac biopsy and by dual photon absorptiometry of the lumbar spine. Bone 9:209–213., , , , ,
- 381985 Correlation between the compressive strength of iliac and vertebral trabecular bone in normal individuals. Bone 6:291–295., ,
- 391989 Relationships among cancellous bone structure and strength in the iliac crest and the lumbar spine. J Bone Miner Res 4:s191 (a294)., , , , , ,
- 401988 Trabecular architecture in iliac crest bone biopsies: Intra-individual variability in structural parameters and changes with age. Bone 9:289–295., , ,
- 411976 Iliac trabecular bone volume in 236 controls: Representativeness of iliac samples. In: JaworskiZGF (ed.) Proceedings of the First Workshop on Bone morphometry. University of Ottawa Press, Ottawa pp. 100–105.,
- 421986 Iliac crest biopsy: Representativity for the amount of mineralised bone. Bone 7:427–430., , , ,
- 431991 A new manual method for assessing two-dimensional cancellous bone structure: Comparison between iliac crest and lumbar vertebra. J Bone Miner Res 6:689–696., , , ,
- 441996 Heterogeneity of the skeleton: Comparison of the trabecular microarchitecture of the spine, the iliac crest, the femur and the calcaneus. J Bone Miner Res 11:36–45., , , , ,
- 451975 Etude de la representativite de la biopsie osseuse iliaque. In: Histomorphometrie de l'osteopenie Physiologique Senile. Thesis Lyon University, Lyon, pp. 89–97., ,
- 461990 Trabecular and endocortical bone remodeling in postmenopausal osteoporosis: Comparison with normal postmenopausal women. Osteoporos Int 1:41–49., , ,
- 471997 Histomoprhometric assessment of the lon-term efects of alendronate on bone quality and remodeling inpatients with osteoporosis. J Clin Invest 100:1475–1480., , , , ,
- 481998 Alendronate increases the mean degree of mineralisation of bone and the uniformity of mineralization of bone in osteoporotic women. (abstract) Bone 23:F282., , , , , , , ,