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



The object of this study was to analyze the cortical thickness (Ct.Th) of the ventral and dorsal shell of the vertebral bodies throughout the human spine in aging and in osteoporosis. Therefore, the complete front column of the spine of 26 autopsy cases (aged 17–90, mean 42 years) without diseases affecting the skeleton and of 11 cases (aged 58–92, mean 77 years) with proven osteoporosis were removed. A sagittal segment prepared through the center of all vertebral bodies was undecalcified, embedded in plastic, ground to a 1 mm thick block, and stained using a modification of the von Kossa method. The analysis included the measurement of the mean cortical thickness of both the ventral and dorsal shell, respectively (from the third cervical to the fifth lumbar vertebral body). The qualitative investigation of the structure of the cortical ring completed the analysis. The presented data revealed a biphasic curve for both the ventral and dorsal shell, skeletally intact with high values of the cortical thickness in the cervical spine (285 μm), and a decrease in the thoracic (244 μm) and an increase in the lumbar spine (290 μm). The mean thickness of the ventral shell is in general greater than the thickness of the dorsal shell in both skeletally normal and osteoporotic cases. The cortical thickness of the spine showed no gender-specific differences (p = NS). There was a slight decrease of the cortical thickness with aging; however, this decrease and the correlation of cortical thickness to age was only significant below vertebral body T8 (r = 0.225–0.574; pr < 0.05–0.005). Most interestingly, however, osteoporosis presents itself with a highly significant loss of cortical thickness throughout the whole spine. This decrease of cortical thickness was more marked in the dorsal shell (p < 0.05) than in the ventral shell (ventral from C3 to T6 [p < 0.05] below T6 [p = NS]). We therefore conclude that in osteoporosis the loss of spinal bone mass is not only a loss of trabecular structure but also a loss of cortical thickness. Furthermore, these results may explain the development of regions of least resistance within the spine in aging and the clustering of osteoporotic fractures in the lower thoracic and lumbar spine.


Osteoporosis is a major generalized bone disease characterized by a low bone mass and the development of nontraumatic fractures, especially of vertebral bodies, as a direct result of osteopenia. There are several types of osteoporosis, the end result of each being a deficient amount of bone mass and bone structure, although different pathogenetic mechanisms may be involved. Since the main region of complications is the spinal column, the decisive organ for human mobility, and at least 538,000 osteoporotic vertebral fractures occur in the U.S.A. each year,1 osteoporosis is of major clinical interest. The histological changes within the spine, especially of the cortical bone, leading to osteoporotic fractures remain poorly understood.

Several morphological investigations demonstrated changes of cancellous bone in aging and due to endocrine dysfunction.2–7 These studies have led us to a better understanding of the cancellous bone structure and its importance for the biomechanical stability of the bone. However, the number of studies investigating cortical thickness (Ct.Th) is remarkably limited. These studies mostly refer to single lumbar vertebral bodies, although more than 50% of osteoporotic vertebral fractures are localized in the thoracic spine.8 The decrease of bone density in aging has been shown to be an important factor in the strength of vertebral bone.9,10 Some other studies reported the importance of the thickness of the cortical shell in the etiology of vertebral fractures.11,12 Thus, the role and contribution of both cortical and trabecular bone to the biomechanical strength of vertebral bodies is still controversial. Furthermore, much of the confusion that exists in the literature on the cortical bone of vertebral bodies is caused by the lack of a clear morphological definition and the structural differences between the cortical ring and the endplates, respectively.

Therefore, the aim of this study was to characterize further the cortical thickness throughout the spinal column and its changes in osteoporosis. Using a systematic histomorphometric approach of direct measurement, we determined the cortical thickness of the ventral and dorsal cortical shell throughout the spine in both normal and osteoporotic subjects. We present evidence of a significant decrease in the cortical thickness in osteoporotics, as compared with normals, which was markedly pronounced in the dorsal shell. Thus, not only trabecular bone mass and structure but also cortical bone seems to contribute to the development of vertebral fracture in osteoporosis.


The complete spine was removed from 37 autopsy cases, composed of 26 healthy controls (13 females and 13 males, aged 17–90 years [mean 42 years]) and 11 osteoporotics (females aged 58–92 years [mean 77 years]). The control group consisted of autopsy specimens who had died in accidents or suddenly without long periods of immobility and without malignant or metabolic bone diseases.

The osteoporotics group were autopsy specimens who had had proven clinical and histological osteoporosis. None of these patients had been taking drugs known to affect the calcium metabolism or had any disease known to cause osteopenia. All of the osteoporotics had fractured vertebrae (from one to six). The fractures were mainly localized in the lower thoracic and upper lumbar spine. Primary and secondary bone diseases were excluded by case history and histological examination of iliac crest biopsies.13

The preparation of the samples was carried out by a standardized procedure as described previously.14,15 Briefly, after removal of the front column of the spine (C3 to L5), a 4 mm thick sagittal segment was prepared with a diamond-coated saw (Fig. 1) and documented on X-rays. Subsequently, the bone marrow was rinsed with water. After dehydration and fat removal, the material was undecalcified and embedded in plastic (methylmethacrylate) according to the method described by Hahn et al.16 The specimens were ground to a thickness of 1 mm, polished, and attached to slides. The surface was stained using a modification of the von Kossa method17 (Fig. 2). The high contrast yield by this silver staining enabled the subsequent evaluation by a semiautomatic image-analyzing system (IBAS 2000, Kontron, München, Germany). The microscopic image was projected onto a digitizing tablet using a video camera (Grundig FA 70 H, Germany). On this image, the thickness of the shell was measured directly. The mean cortical thickness of both the ventral and dorsal shell of each vertebral body was accessed by multiple measurements spaced equidistantly (300 μm, ventral 20–50 times and dorsal 10–30 times for each vertebral body) from C3 to L5 (Fig. 3). The shortest distance between the outer surface and the endocortical surface was determined. The corner regions of the vertebral bodies and the region of the foramen of the central vein of the dorsal shell were excluded to eliminate artifacts (e.g., osteophytes) caused by degenerative changes. All fractured vertebral bodies of the osteoporotic group were omitted from the histomorphometric analysis. Due to this fact, the mean value of some vertebral bodies (T6, T10, T12, and L2) of the osteoporotics represents only 8–10 specimens, respectively. Using this method, the intraobserver variation for repeated measurements of a single vertebral body shell was ±15 μm and the interobserver variation between two experienced users was ±25 μm.

Figure FIG. 1.

The drawing indicates precisely where the 4-mm sections were taken from the vertebrae.

Figure FIG. 2.

Representative embedded specimens of the vertical section of two lumbar vertebral bodies. Transmission light illumination, von Kossa surface-stained 1 mm thick blockgrinding. Scalebar = 5 mm.

Figure FIG. 3.

Diagram of the measurement of the cortical thickness of the ventral and the dorsal shell. The mean value of each shell of one vertebral body is represented by multiple measurements of the cortex.

Statistical analysis

The mean values of the cortical thickness (ventral and dorsal for each vertebral body) of the controls were compared with the osteoporotics and also by age and gender. The significance of differences between mean values was assessed by the Mann-Whitney U-test. Spearman's rank correlation coefficient was used to test relationships between variables. The value p < 0.05 was chosen as the significance level.


Controls and osteoporotics

The mean vertebral cortical thickness of both the skeletal intact specimens (controls) and osteoporotics, showed a biphasic curve with high values in the cervical spine, a decrease in the thoracic spine, and again an increase in the lumbar spine (Figs. 4A and 4B). The mean cortical thickness of the controls was, in the cervical spine, 329 μm for the ventral shell and 240 μm for the dorsal shell; in thoracic spine, 268 μm for the ventral shell and 220 μm for the dorsal shell; and, in the lumbar spine, 308 μm for the ventral shell and 272 μm for the dorsal shell. In contrast, the mean cortical thickness of the osteoporotics of the cervical spine was 238 μm for the ventral shell and 176 μm for the dorsal shell, in the thoracic spine; 238 μm in the ventral shell and 164 μm in the dorsal shell; and, in the lumbar spine, 321 μm in the ventral shell and 185 μm in the dorsal shell, respectively.

In osteoporosis, the ventral cortical thickness was on average about 15% lower than in the controls. While this difference was found to be significant in the cervical and upper thoracic spine (p < 0.05), in the lower thoracic and lumbar spine the reduction of the ventral cortical thickness in osteoporosis as compared with controls was not statistically significant (p = NS) (Fig. 4A). In sharp contrast, the mean values of the dorsal shell were significantly (p < 0.05 except T6) reduced throughout the spine as compared with the controls (on average about 30%) (Fig. 4B). The same applies for the comparison between controls and osteoporotics for the total mean value of each of the vertebral bodies.

Figure FIG. 4.

Cortical thickness of the (A) ventral shell and of the (B) dorsal shell of the vertebrae of both controls and osteoporotics, respectively. Mann-Whitney U-test was used and the standard error was given.

Within the controls (male, mean age 40 years, and female, mean age 43 years) the cortical thickness did not depend on gender for both the ventral and dorsal shell (p = NS) (data not shown). Furthermore, there was no significant age-related decrease for the total mean value (amount of the ventral and dorsal) of each vertebral body in the region C3 to T7 (r = 0.024–0.362, pr = NS) (Fig. 5A). However, a significant age-related decrease in the cortical thickness was observed below the vertebral body T8 (r = 0.225 to 0.574, pr < 0.05–0.005) (Fig. 5B).

Figure FIG. 5.

Age-dependent loss of cortical thickness of the vertebral bodies. (A) The region C3 to T7, represented by the vertebral bodies C5 and T5, showed a decrease with age but this decrease was not significant (r = 0.024–0.362, pr = ns). (B) The decrease in cortical thickness was significantly correlated with age below T8, represented here by the vertebral bodies T10 and L3 (r = 0.225–0.574, pr < 0.05–0.005).

Qualitative results

In contrast to the thick and compact cortex of the long bones, the cortex of vertebral bodies consist of a thin and inhomogeneous structure (Figs. 6A and 6B). The typical cylindrical units (referred to as haversian systems or osteons) characteristic for cortical bone are markedly less apparent in the vertebral cortex as compared with the cortex of the long bones (Fig. 7). The observed differences between the ventral and dorsal shell, the different regions of the spine (cervical, thoracic, and lumbar), and between controls and osteoporotics, respectively, were always quantitative differences rather than qualitative structural differences.

Figure FIG. 6.

(A) Sagittal slice of the ventral shell and (B) of the dorsal shell from a representative vertebral body in dark light illumination revealing the three-dimensional quality of the specimens. The black area demonstrates the surface stained bone, while the white area is unstained bone below the surface (the block grindings are 1 mm thick). The anterior and posterior longitudinal ligaments are indicated by asterisks, the cortical bone by arrows, and the trabecular bone by arrowheads (von Kossa staining, 1 mm thick blockgrinding. Scalebar = 1 mm).

Figure FIG. 7.

Horizontal slice of the ventral shell in transmission light illumination highlighting the surface and resembling a two-dimensional histological section. Osteons appear as holes (arrows) which are a hallmark for cortical bone (von Kossa surface-stained 1 mm thick block grinding). Scalebar = 1 mm.


Vertebral bodies consist of two main structural components: trabecular and cortical bone. Several histomorphometric studies of the vertebral trabecular microarchitecture of skeletally healthy patients showed an age-related loss of the trabecular bone structure.18–20 Furthermore, osteoporosis was characterized histomorphologically by a rapid loss of trabecular bone mass, a removal of complete structural elements, and reduced trabecular interconnection. As a result, local monostotic as well as polyostotic differences throughout the spine were observed and gained importance for the etiology of osteoporotic fractures.21,22 These investigations concluded that not only bone mass but also the trabecular bone structure is an important factor for biomechanical stability.

The relative contribution of cortical versus trabecular bone to whole bone strength is, however, still controversial23 and remains poorly understood. The majority of osteoporotic fractures occur in the lower thoracic spine; however, the limited number of related studies mainly focused on the lumbar spine. Rockoff et al.24 found that the cortex of lumbar vertebral bodies contributes 45–75% to the peak strength, while in contrast, McBromm et al.25 found a reduction in strength of only approximately 10% when the cortex was removed. Differences in preparation techniques of the vertebral bodies, in cortical segmentation, and in loading technique might explain the different results of these studies. Yoganandan et al.26 and Mosekilde27 found nearly the same results (22–54% and 26–57%, respectively) as the contribution of the cortex to the total vertebral strength.

Much of the confusion that exists about the role of the vertebral cortex is due to the comparison of biomechanical results with indirect measurements of the cortical thickness because systematic, direct histomorphometry accessed values for the cortical thickness are not available. The direct measurement of the cortical thickness on midsagittal grinding samples, however, offers several advantages compared with any other method. The histomorphometric measurement on undecalcified samples gives access to most accurate data by the elimination of possible artifacts and avoiding overlapping phenomena, which often occur in CT-based measurement, thereby leading to an overestimation of the cortical thickness. Furthermore, this histological approach clearly demonstrates the structural differences between the ventral and dorsal shell on one side and the endplates on the other side. Although the whole bony shell of vertebral bodies has been described as condensed trabecular bone by some authors,28 the proof of osteons in both the ventral and dorsal shell clearly mark them as cortical bone structures.

Most interestingly, the cortical thickness shows a distribution pattern throughout the spine that is very similar to that reported for BV/TV.29 Thus, regional differences of the trabecular bone within the spine go along with a parallel behavior of the cortical thickness. However, the polyostotic heterogeneity described in osteoporosis21 is a heterogeneity of trabecular bone rather than one of the cortical bone. Nevertheless, osteoporosis leads to significant changes in cortical thickness of the vertebral bodies. Throughout the spine, the thickness of the dorsal shell is significantly reduced as compared with normal controls. Within the cervical and thoracic spine also the ventral shell was significantly thinner, while for the latter this was not observed in the lower thoracic and lumbar spine. The mechanisms by which cortical bone loss is prevented in the lower thoracic and lumbar spine remain to be determined. One has to take into account, however, that this observation could be due to the study design. Since the mean values were calculated only from intact vertebral bodies, it is possible that the fractured vertebral bodies were comprised of vertebral bodies with a thinner cortex, thereby increasing mean values. Throughout the spine, the ventral shell is in general thicker than the corresponding dorsal shell in both normal controls and osteoporotics, respectively. The disruption of the dorsal cortex by the foramen of the basivertebral vein and differences in biomechanical loading might be possible explanations, as well as the differential weight-bearing on the anterior and posterior cortices at different points along the S-shaped spine.

Since the ventral and dorsal shell showed the same age-dependent behavior, the correlation of the cortical thickness to age was done using the total mean cortical thickness per vertebral body. There is no significant age-related decrease of the cortical thickness between C3 and T7 (r = 0.024–0.362, pr = NS); in contrast, however, both parameters were significantly correlated between T8 and L5 (r = 0.225–0.574, pr < 0.05–0.005). The reasons for this finding remain poorly understood. Although it is possible that this result is due to the limited sample size, there are other alternatives to this interpretation, for example region-dependent structural differences as known for trabecular bone. In agreement with previous studies,30 the cortical thickness is independent of gender. In contrast to the findings described in the text, Silva et al.30 reported that the cortical thickness of the first lumbar vertebrae is age independent. An obvious limitation of the latter study was, however, that out of his 16 specimens the youngest was 43 years old.

Our results further confirm previous studies by Vesterby et al.12 (measuring 7 μm thick undecalcified histological sections) and Jayasinghe et al.31 (three-dimensional photographic study) with respect to the age-dependent loss of cortical bone within the lumbar spine. These studies as well as other radiographic studies32,33 do also not contradict our findings within the cervical and thoracic spine, since all of them were limited to vertebrae within the lumbar spine.

The data reported on the cortical thickness of the vertebral bodies are not uniform but rather show a widespread range. In close accordance with our results are the studies of Whitehouse et al.,35 Silva et al.,30 and Vesterby et al.12 Based on scanning electron microscopy of one vertebral body, Whitehouse et al.35 reported values between 200 and 250 μm for the cortical shell. Using methods similar to ours, Silva et al.30 found mean values ranging from 180 to 600 μm, and Vesterby et al.12 found values of L1 ranging from 220 to 600 μm (mean 410 μm). In contrast, Arnold34 found in young adults values from 300 to 1000 μm measuring the cortex on X-rays without describing values for older adults.

Ma et al.36 even reported much higher mean values ranging from 1620 to 5190 μm with a thicker anterior than posterior shell and an increase from thoracic to lumbar vertebral bodies; however, they described neither the used material nor the methods accurately.

Thus, the present study provides new data on the cortical thickness and its changes in aging and due to osteoporosis throughout the human spine. These results should be complementary to the histomorphometric data of trabecular bone, which are available in the literature. We therefore conclude that (1) the cortical thickness of the spinal column is characterized by a biphasic curve with peak values in the cervical and lumbar spine but lower values for the thoracic spine in both ventral and dorsal shell, respectively, (2) the mean thickness of the ventral shell is significantly higher than that of the dorsal shell, (3) there is an age-related decrease in cortical thickness, which is significant in the lower thoracic and lumbar spine, (4) cortical thickness is the same in males and females, and (5) osteoporosis leads to a loss of cortical thickness, which is markedly pronounced in the dorsal shell. This suggests that, in addition to trabecular bone measurements, the latter region might be of special interest for diagnostic radiological examinations.


The authors would like to express their thanks to Jutta Rieck for excellent photographic assistance. M.A. is a fellow of the German research community (DFG Am 103/2-1). This study was supported by the Berufsgenossenschaft Hamburg.