Bone Marrow Composition and Bone Microarchitecture and Turnover in Blacks and Whites


  • Christine M. Schnitzler,

    Corresponding author
    1. MRC Mineral Metabolism Research Unit and Department of Orthopaedic Surgery, University of the Witwatersrand, Johannesburg, South Africa
    • C.M. Schnitzler, 608 Helderberg Village, Private Bag X19, Somerset West 7129, South Africa
    Search for more papers by this author
  • Julia Mesquita

    1. MRC Mineral Metabolism Research Unit and Department of Orthopaedic Surgery, University of the Witwatersrand, Johannesburg, South Africa
    Search for more papers by this author


We examined the relationship between bone histomorphometric variables versus marrow cellularity, marrow adiposity (among hemopoietic cells), and fatty degeneration (areas of only fat) of bone marrow in iliac crest bone samples from 98 normal black (n = 53) and white (n = 45) males and females. We found blacks to have greater marrow cellularity (p = 0.0001), less marrow adiposity (among hemopoietic cells, p = 0.0001), greater values for bone volume (p = 0.030), trabecular thickness (p = 0.002), and static bone turnover variables (osteoid volume, p = 0.001; osteoid surface, p = 0.001; osteoid thickness, p = 0.001; eroded surface, p = 0.0006) than whites. Marrow cellularity correlated positively with static bone turnover variables osteoid volume (r = 0.257, p = 0.011), osteoid surface (r = 0.265, p = 0.008), osteoid thickness (r = 0.217, p = 0.032), and eroded surface (r = 0.273, p = 0.007) when all 98 cases were analyzed together. These findings suggest that marrow cells may influence bone turnover. The extent of fatty degeneration, but not that of adipose tissue, increased with age in blacks (r = 0.476, p = 0.0003) and whites (r = 0.476, p = 0.001), as did bone loss. There was no racial difference in the extent of fatty degeneration. We conclude that the lesser extent of adiposity in blacks is a racial characteristic that is unaffected by aging, whereas fatty degeneration which may have partly occupied space vacated by bone loss, is an aging phenomenon, unrelated to race. Greater bone turnover in blacks may be expected to lead to more frequent renewal of fatigue-damaged bone, which together with sturdier bone structure may contribute to the lower fragility fracture rates in blacks.


TRABECULAR BONE ALWAYS COEXISTS with bone marrow. This applies not only to orthotopic marrow sites in flat and cuboidal bones and the metaphyses of long bones but also to heterotopic sites. Bone marrow forms not only in new trabecular bone growing from existing bone (osteoconduction) but also in trabecular bone developing de novo within soft tissue (osteoinduction).1 This intimate relationship suggests an interaction between marrow and bone. Indeed, there now exists ample evidence for the existence of bone marrow-derived cells that are able to differentiate into either osteoblasts or adipocytes.2–6 Furthermore, previous studies have reported positive correlations between marrow cellularity and bone volume (BV),7 and some but not all investigators found correlations between marrow cellularity and bone turnover.8–11 Moreover, marrow adiposity has been shown to increase with declining bone mass.7

Previous histomorphometric examination of iliac crest bone samples from 346 normal black and white South Africans revealed that blacks had thicker trabeculae, greater BV (males only), and greater values for static bone turnover variables.12 In the course of that study, greater bone marrow cellularity was noticed in some specimens than in others, and some showed areas of fatty degeneration in the form of patches of almost exclusively fat cells. These observations prompted a re-examination of these bone specimens to establish any possible relationships between bone marrow features and bone histomorphometric variables.



Ninety-eight adult subjects, 53 blacks (28 male, 25 female) aged 40 (25–56) years and 45 whites (23 males, 22 females) aged 41 (25–57) years were studied. Values are given as median (10th–90th percentile), unless otherwise indicated. Six of the 98 subjects were patients who consented to a transiliac bone biopsy on the occasion of a minor operation on the periphery of an upper or lower limb; all were walking unaided before biopsy. The other 92 subjects were previously healthy individuals who had died suddenly and who had no evidence of organic disease at autopsy.12

Bone samples

Transiliac bone cores, 7.5 mm in diameter, consisting of the outer and inner cortices and the intervening cancellous bone were obtained from the standard site, namely 2 cm below the iliac crest and 2 cm behind the anterior superior iliac spine.13 The specimens were processed undecalcified and analyzed for microstructural and static bone turnover variables by routine histomorphometry as previously described.12 Nomenclature and definitions of measured and calculated variables are those approved by the American Society for Bone and Mineral Research.14 All 98 specimens had been analyzed by routine histomorphometry in a study of 346 cases aimed at establishing normal reference values for histomorphometry for South African black and white adults.12 The 98 specimens were selected from the 346 to compile four to six age- and gender-matched black/white pairs for each 5-year age group. Specimens with suboptimal preservation of cellular detail of bone marrow were excluded.

Marrow fat was measured as two separate variables: adiposity volume (AdV) and fatty degeneration (FaDg). Adiposity was the adipocytes interspersed among hemopoietic cells. The terms adipocyte and adiposity will hereafter only be used to describe fat cells among hemopoietic cells. FaDg describes unevenly distributed accumulations of almost exclusively fat cells, varying in extent from clusters of 15 or more cells (Figs. 1a and 1b) to replacement by only fat cells of whole marrow compartments between neighboring trabeculae. Marrow adiposity and marrow cellularity were measured in the central portion of cancellous bone, thus avoiding transitional bone at the cortico-cancellous junction. A Zeiss Integrationsplatte II (Zeiss, Jena, Germany; 100 grid points where horizontal and vertical lines cross) was placed in one eyepiece, superimposing it on the specimen at a total magnification of ×100.

Figure FIG. 1.

Photomicrograph and corresponding diagram of iliac crest bone from a 45-year-old black male who died in an accident; autopsy revealed no organic disease. (a) Photomicrograph. The trabecula is surrounded by fat cells; hemopoietic marrow is situated away from bone. A small area of FaDg consisting exclusively of fat cells is seen at the bottom left. Original magnification ×10. (b) Diagram of the same photomicrograph. BV, bone volume; HmMa, hemopoietic marrow; Ad, adipocytes among hemopoietic cells; Ce, hemopoietic cells; FaDg, fatty degeneration.

Sections were scanned in rectilinear fashion and all hits (grid points) on marrow adipocytes and on cellular (hemopoietic) marrow were recorded until 1000 hits were reached per specimen. Trabecular bone and areas of FaDg were excluded from this measurement, but blood vessels were included under cellular marrow. The percentage of cell volume in hemopoietic marrow volume (HmMaV) was calculated as: CeV/HmMaV% = N hits cells/(N hits cells + N hits adipocytes) × 100. The percentage of adipocyte volume in HmMaV was calculated as: AdV/HmMaV% = N hits adipocytes/(N hits cells + N hits adipocytes) × 100.

FaDg was measured separately because it often extended beyond the central portion of cancellous bone where the other variables were assessed. FaDg was measured semiquantitatively as follows: 0 = none, 1 = few, 2 = many, and 3 = widespread areas of almost exclusively fat cells. The measurements of cellularity and adiposity using referent HmMaV instead of referent tissue volume (TV), and measurement of FaDg in semiquantitative grades permitted quantitation of these variables without the confounding effect of changeable components (e.g., bone volume, BV/TV) within the referent TV. However, for calculations relating bone tissue components to each other, all marrow variables had to be expressed by referent TV. For variable FaDg volume in tissue volume (FaDgV/TV) this required measurement by point counting: FaDgV/TV% = N hits FaDg × 100/N fields × 100. The following conversions were carried out for the other two marrow variables: cell volume in tissue volume CeV/TV% = (100 − BV/TV-FaDgV/TV) × CeV/HmMaV/100), and adipose volume in tissue volume AdV/TV% = (100 − BV/TV − FaDgV/TV) × (AdV/HmMaV/100).

Statistical analysis

The Statistical Analysis System (SAS Institute, Cary, NC, U.S.A.) was used throughout. Differences in marrow composition, age and histomorphometric variables between blacks and whites and between the genders were tested with the rank sum test. Differences in fatpatch extent between the races and genders were tested with the chi-square test (Fisher exact test). Correlations between age and the histomorphometric variables versus marrow cellularity and marrow adiposity were tested with the Pearson correlation coefficient, and those versus extent of FaDg were tested with the Spearman rank correlation coefficient. Multiple regression analysis for maximum r2 was used to examine the contribution made by age, marrow cellularity, marrow adiposity, and extent of FaDg to the variance of histomorphometric variables.


Marrow composition versus race, gender, and age

Values for marrow cellularity (CeV/HmMaV) were greater, and those for marrow adiposity (AdV/HmMaV) lower in blacks than in whites (Table 1 and Fig. 2). This also applied when males and females were tested separately (males: CeV/HmMaV 64.2 [51.1–77.6] vs. 50.7 [34.6–65.8], p = 0.0001; AdV/HmMaV 35.9 [22.4–48.9] vs. 49.3 [34.2–65.4], p = 0.0001; females: CeV/HmMaV 67.2 [54.6–78.2] vs. 54.4 [42.2–69], p = 0.0004; AdV/HmMaV 32.8 [21.8–45.4] vs. 45.7 [31–57.8] p = 0.0004). There were no significant gender differences for CeV/HmMaV or AdV/HmMaV in either race, and neither were there any significant correlations between CeV/HmMaV or AdV/HmMaV versus age (Table 2). FaDg, however, showed no significant race (Table 1) or gender differences, but it correlated positively with age in both race groups (Table 2 and Fig. 3). FaDg correlated negatively with CeV/HmMaV in both races (Table 2).

original image
original image
Figure FIG. 2.

Histogram to show the difference in bone marrow cellularity in normal black (n = 53) and white (n = 45) adults.

Figure FIG. 3.

Graphs to show the effects of age on bone tissue components BV, adiposity, FaDg, and hemopoietic cellularity in iliac crest bone samples from normal black (n = 53) and white (n = 45) adults.

Bone histomorphometry versus race, gender, and age

Blacks had greater values for BV/TV, trabecular thickness (Tb.Th), and static bone turnover variables than whites (Table 1). The only gender difference was greater osteoid volume in black males (2.76% [0.93–4.9%]) compared with black females (1.41% [0.29–4.89%], p = 0.035). Since this was the only gender difference in the study, males and females were analyzed together. Age had a negative effect on bone structure and in blacks correlated positively with the static bone turnover variables osteoid volume, osteoid surface, and eroded surface (Table 2).

Marrow composition versus bone histomorphometry

Blacks had greater BV/TV, more marrow cellularity (CeV/HmMaV), and less marrow adiposity (AdV/HmMaV) than whites (Table 1). The same applied when young (less than 40 years) and old (40 years and above) subgroups were tested separately (Table 3). Fatty deneration (FaDg), however, showed no racial difference (Table 4) nor within the young and old subgroups (Table 3). As bone shrank with aging from the third decade onward, FaDg expanded, but not until some years later, and not to the same extent as bone was lost (Fig. 3); the remaining space was taken up by a slight increase in CeV/HmMaV (Table 3). AdV/HmMaV did not increase with age. Changes of aging were similar in blacks and whites. CeV/HmMaV correlated negatively with FaDg in both blacks and whites (Table 2), and it correlated positively with static bone turnover variables when blacks and whites were analyzed together (OV/BV, r = 0.257, p = 0.011; OS/BS, r = 0.265, p = 0.008; O.Th, r = 0.217, p = 0.032; ES/BS, r = 0.273, p = 0.007). In blacks, FaDg correlated negatively with BV/TV, Tb.Th, and trabecular number (Tb.N), and positively with trabecular separation (Tb.Sp) (Table 2). Multiple regression analysis (only r2 values above 9% are given) showed that FaDg was the most frequently recurring independent variable accounting significantly for the variance of structural bone variables, and age the most frequently recurring independent variable accounting significantly for the variance of static bone turnover variables (Table 5).

original image
original image
original image


This study has shown that bone marrow in blacks contains a greater proportion of hemopoietic cells and less fat than in whites, and that with aging, areas of FaDg appear in the marrow in both races. It has also repeated our previous findings in 346 normal black and white individuals12 of greater values in blacks for BV/TV, Tb.Th, and static bone turnover variables osteoid volume, osteoid surface, osteoid thickness, and eroded surface. The positive correlations between bone marrow cellularity and static bone turnover variables obtained when all 98 cases were analyzed together suggest an interaction between bone turnover and marrow cellularity. This interaction has been shown to be mediated by growth factors and cytokines.3,15–19 These substances control in particular the differentiation of bipotential stromal cells to either osteoblasts or adipocytes. Aging too was found to affect this differentiation switch, namely through differential gene expression in osteoblast progenitor cells.6 Another study20 found a decline in numbers of capillaries and sinusoids to accompany the fatty replacement of marrow and the decline in BV/TV. The investigators of that study also describe intracytoplasmic accumulations of glycoprotein containing droplets in endothelial cells of paratrabecular but not of central marrow sinusoids; the extent of these accumulations correlated with osteoblastic osteoid deposition. This finding assigns a pivotal role to paratrabecular endothelium in bone turnover and maintenance of bone mass. Structural characteristics of marrow vascularity also appear to relate to marrow composition since large vascular sinusoids with discontinuous endothelial lining were found in hemopoietic marrow, and a closed capillary bed in fatty marrow.21 The link between vascularity and osteoblast differentiation and proliferation may also be local growth factors and cytokines that may in turn be hormonally and physically controlled.16,22 It remains unclear though what determines changes in vascularity.

The importance of the finding of greater marrow cellularity and possibly greater bone turnover in blacks lies in the influence on skeletal competence. Among its functions, bone turnover is responsible for the removal of fatigue-damaged bone and its replacement with new bone. With greater bone turnover, bone will be renewed more frequently, will accumulate fewer loading cycles, and be less prone to fatigue failure and fragility fractures. This beneficial effect would of course only pertain if there is balance in the remodeling cycle so that bone mass is maintained. The cause for greater marrow cellularity in blacks remains unknown.

It has been suggested that fatty replacement of marrow precedes bone atrophy. However, the finding in this study of a time lag of several years between age-related bone loss and increase in FaDg suggests that bone atrophy precedes FaDg. Also, suggesting a bone-loss-first sequence is the striking histologic picture of some early osteoporoses in which fat cells surround trabeculae while hemopoietic tissue remains located deeper in the marrow space, away from bone20 (Fig. 1). Such fat-lined bone surfaces usually lack bone turnover activity. These appearances raise the question of whether bone sends—or fails to send—messages to nearby marrow stromal cells which then preferentially differentiate into adipocytes rather than into osteoblasts.3,4,23 Such a message could come from advanced glycation end products derived from aging collagen which have been shown to suppress osteoblastic differentiation and function.24 Alternatively, these rows of fat cells could result from a local differentiation switch to adipogenesis and simply be replacing bone where thinning trabeculae have vacated space. Furthermore, resetting of the skeletal mechanostat25,26 could conceivably have contributed to fatty change in marrow adjacent to bone. Our finding of bone atrophy preceding FaDg by several years also suggests that bone atrophy may have led to FaDg and not FaDg to bone atrophy. Other investigators27 too found, in a study of rats, that marrow fat increased after BV/TV had begun to decline.

Areas of FaDg may be the result of a localized switch from osteogenesis to adipogenesis which is presumably a degenerative phenomenon of aging since FaDg increased with age, as did structural deterioration of bone. It is tempting to assume that fatty degeneration filled the space vacated by bone loss since on average the extent of bone loss and that of increase in FaDg were not too dissimilar: about 60% of bone space vacated was taken up by FaDg (Table 3), although a cross-sectional study such as this cannot provide exact figures. The fat-for-bone hypothesis is also supported by the results of multiple regression analysis: FaDg came up most frequently as a statistically significant independent variable accounting for the variance of structural variables of bone (Table 5). However, the large size of many areas of FaDg compared with that of trabecular structures argues against a simple fat-for-bone exchange of space. Moreover, many areas of FaDg were still traversed by trabeculae which they were supposed to have replaced.

The space occupied by greater BV/TV and marrow cellularity in blacks was compensated for by a lower AdV than in whites (Table 4). This difference already existed in the under 40 years age group (Table 3) and was therefore not related to aging. The space vacated by the bone loss of aging, however, was in both races made up, in great part, by an increase in FaDg and not by any increase in adiposity; this feature was similar in blacks and whites. It would appear therefore that the extent of adiposity among hemopoietic cells is a racial characteristic, unaffected by aging, whereas FaDg is a characteristic of aging, irrespective of race.

Other workers studying age-related changes in iliac crest bone marrow7 presumably included both, adiposity among hemopoietic cells as well as areas of FaDg, in the measurement of marrow fat and so found an age-related increase in marrow fat; we found an age-related increase only for FaDg. This finding obviously applies only to iliac crest bone in the age groups examined and not to the skeleton as a whole because of regional variations in marrow composition.28

We conclude that, first, blacks had greater marrow cellularity, less marrow adiposity (among hemopoietic cells), more bone, thicker trabeculae, and greater values for static bone turnover variables. Marrow cellularity and adiposity did not change with age. Second, areas of FaDg of marrow were of similar extent and increased similarly with age in blacks and whites; bone loss preceded this increase. Third, extent of adiposity appears to be a racial characteristic, unaffected by aging, and FaDg a characteristic of aging, irrespective of race. Fourth, if greater values for static bone turnover variables in blacks reflect greater bone turnover, then more frequent renewal of fatigue-damaged bone may be expected to contribute, together with sturdier bone structure, to the lower fragility fracture rates in blacks.


We thank Coral Gordon for typing the manuscript and the staff of the Photo Illustration Unit for preparing the figures and tables. This study was funded by the Medical Research Council of South Africa, and by the University of the Witwatersrand, Johannesburg, South Africa.