Department of Endocrinology and Medicine, Austin and Repatriation Medical Centre, University of Melbourne, 3084 Melbourne, Australia
Address all correspondence and requests for reprints to: Ego Seeman, B.Sc., MBBS, FRACP, M.D., Department of Endocrinology, Austin and Repatriation Medical Center, Heidelberg 3084, Melbourne, Australia.
Where is the wisdom we have lost in knowledge? Where is the knowledge we have lost in information?
From The Rock by T. S. Eliot
Techniques are not ends in themselves, they are only as good as the answers they provide. The answers are only as good as the questions, and the questions are only as good as the insights generated from observations made using the technique. Problems arise when the technique takes over, like HAL in Stanley Kubricks 2001: A Space Odyssey, driving inquiry because of the accessibility of its "ON" button, but misdirecting inquiry by producing flawed inferences due to the techniques unrecognized limitations (1, 2).
Although the bone densitometer has a central role in the diagnosis, prevention, and treatment of osteoporosis in clinical practice, it has taken over in clinical research ("Im sorry Dave, ... . . I cant do that... . "), and its printout, the bone "density" measurement, has become an end in itself, producing conceptual errors because of its technical limitations.
The use of densitometry has led to several misconceptions: 1) volumetric bone mineral density (BMD) increases during growth. It does not. Growth builds a bigger, not more dense, skeleton; 2) peak volumetric BMD is higher in men than women. It is not. Bone size is greater; 3) women lose more trabecular bone from the spine than men. No so, there are qualitative, not quantitative, differences; 4) cortical bone loss is less in men than women during aging. This is not so; the amount of bone resorbed from the endocortical (inner) surface of the cortical shell is similar in men and women. The amount of bone formed on the periosteal (outer) surface during aging is greater in men so that "net" cortical bone loss is less in men; and 5) bone mass is reduced in patients with fractures. It certainly is, but the deficit is exaggerated in patients with smaller bones than controls and underestimated in patients with larger bones than controls. These erroneous notions are produced by uncritical use of densitometry.
Density and confusion
Bone consists of the mineralized matrix fashioned into a cortical
shell and a trabecular network of plates within a marrow cavity. Only
the mineralized mass of bone is seen by the photons so that the degree
of attenuation of the photons during their transmission through the
bone is a measure of the bone mineral content (BMC) of the whole
bone, not all of which is mineralized tissue. The image produced during
scanning is like a silhouette, just a two-dimension image of the
three-dimensional structure, an "areal" projection of the bone in
the coronal plane (Fig. 1A). The length
and width of the scanned bone is known, but not its depth. Because the
depth of the bone is not seen, a bone with greater depth will attenuate
more photons and will be reported as being more dense. The bone is not
more dense, it does not have a greater amount of bone mineral
distributed (as thicker trabeculae, more trabeculae, a thicker cortex)
within the periosteal envelope, the bone is just larger; there is more
mineral, but this mineral is distributed in the larger volume of the
whole bone.
|
The areal BMD measurement is a function of the size of the bone and the
amount of bone within its periosteal envelope; its volumetric BMD. The
problems arise because bone size and the amount of bone within it
change independently during growth and aging, and often in opposite
directions. For example, during growth, external bone size increases
with a proportional increase in the amount of bone within its
periosteal envelope so that the volumetric BMD remains constant, but
areal BMD increases because the bone is bigger. During aging, the
amount of bone within the periosteal envelope decreases as bone
resorption results in thinning and loss of trabeculae, thinning, and
increased porosity of the cortex (Fig. 1B).
However, aging is also associated with simultaneously occurring periosteal apposition, which enlarges the bone and offsets resorptive removal of the mineralized matrix (3, 4, 5). Consequently, the net loss of bone (the sum of bone deposit on the outside and removed on the inside) underestimates the absolute bone lost by resorptive removal inside the bone. Patients with spine fractures have smaller bones than controls (6). Because the areal BMD measurement only sees less bone mineral mass without distinguishing whether its due to a smaller bone, less bone in the bone or both, the lower areal BMD in patients with fractures overestimates the deficit in bone mass inside the smaller bone (produced by excessive bone loss or reduced accrual). The densitometric measurement does not see these opposing surface-specific changes but rather integrates their net effects.
The purpose of this commentary is to describe the structural and biomechanical basis of: 1) the greater gain bone strength during growth in men than in women; 2) the lesser decline in bone strength in men than in women during aging; and 3) bone fragility in patients with fractures drawing attention to the conceptual errors produced by densitometry in the clinical context of skeletal growth and aging.
Growth-designing structure for function
Bone mass increases during growth because bone length and diameter increase. The mass of bone contained within the periosteal envelope of the bone also increases. The growing bone is fashioned into a structure adapted to carry out specific functions. The vertebral bodies consist of an interconnecting trabecular network of plates and sheets forming a marrow filled spongiosa, a mass of cancellous bone contained within a thin cortical shell.
The design secrets of weight-bearing structures were known by the Ancients and are still standing despite thousands of years of gravity. Long bones like the radius and femur grow in length and diameter. The mineralized bone mass within their periosteal envelope is not formed a solid column that will easily tolerate loads but will remain immovable no matter how hungry the owner, or how delicious the meal scurrying by. The mineralized mass of bone within the periosteal envelope of the long bones is modeled into hollow cylinders with splayed ends containing trabecular bone. These structures are architectural masterpieces of minimalism, created using the least amount of material needed for optimum functionlightness for speed, strength for loadingensuring the organism can find dinner, rather than be dinner.
Growth fashions a bigger, not denser, skeleton. BMD increases
during the pre- and peripubertal years because long bones increase in
size (Fig. 2, top)
(7). The densitometer sees a bigger bone and prints out
that the bigger bone has a higher areal BMD, a higher density, giving
the impression that as children grow their bones have somehow become
more dense. Similarly, in late puberty, when boys have long bones of
greater diameter than girls, the densitometer sees a bigger bone and
prints out a higher areal BMD in boys, giving the impression that
somehow boys have grown denser bones than girls. This is not so.
|
Volumetric BMD also remains independent of age for the vertebral body,
at least until puberty (10). In this case, the mass of
bone is fashioned as a trabecular network. The length and thickness of
the intersecting trabecular sheets increases in proportion to the
enlarging medullary space containing them. The number of trabeculae,
established at the growth plates, do not increase with age
(11). At puberty, trabecular BMD increases and does so to
a similar degree in boys and girls (Fig. 3) (10), perhaps
because trabeculae continue to thicken while growth in external size of
the vertebral body ceases. Growth in external size ceases sooner than
the accrual of bone within its periosteal envelope
(12, 13, 14). However, longitudinal studies documenting the
precise temporal relationship of cessation of growth in external size
and accrual of mass within the vertebral body are lacking.
|
The position of volumetric BMD in normal distribution is determined before birth. The constancy of volumetric BMD before puberty has very important implications. First, if volumetric BMD is independent of age, then the position of an individuals volumetric BMD must be determined before birth. Second, volumetric BMD of the spine probably tracks (remains in the same position in the population distribution as age advances) (15). Volumetric BMD is a function of the relative growth in size of the whole bone and the mass accrued within it, not the absolute growth in size or mass. Thus, there must be regulators and coregulators of bone size and mineral accrual that ensure that growth of size is matched precisely with a given proportional growth in mass within it.
Consequently, a persons volumetric BMD at axial or appendicular regions tracks through growth maintaining the same position in the normal distribution at 18 yr of age as was present at 2 yr of age. This tracking is well documented in the pediatric literature for height, weight, skull size, and many other traits. Third, for there to be variation in volumetric BMD, some individuals must accrue more bone per unit external volume of bone than others to be placed at the 95th percentile, whereas others must accrue less bone per unit external volume of bone to be placed at the 5th percentile (i.e. some individuals must have less cortical thickness, fewer or thinner trabeculae per unit external bone volume than others, placing them at greater risk for fragility fractures during aging).
What genetic, hormonal, biomechanical and other factors account for the variance in volumetric BMD? What determines the relationship between the growth of the skeleton in size and the mass within it? What regulates the variance in shaft width, cortical thickness, trabecular number for a given bone length? We will not know until we define the age- and gender-specific variance of these structural components of areal BMD (not areal BMD) and then study the candidate genes or environmental factors that account for components of the variance in each of these structural components (not areal BMD).
The absolute and relative movement of the periosteal and endosteal surfaces determine bone size and its mass. The gender difference in leg length is the result of a longer prepubertal period of growth in the male (16). During this time, growth is more rapid in the legs than spine (14). The pubertal growth spurt occurs 1 yr later in boys, and pubertal growth velocity is greater and more protracted than in girls. At puberty, growth velocity of the appendicular skeleton slows and then ceases whereas growth velocity of the spine accelerates to exceed that of the legs and continues longer. The difference in height at peak between women and men is primarily in leg length, not trunk length. (Spouses should negotiate sitting, not standing.) The long bones and vertebrae are also wider, but the vertebrae are only slightly taller in men than women.
The absolute and relative movements of the periosteal and endosteal surfaces determine the diameter of the long bone, the mass of cortical bone, cortical thickness, and the distance this cortical mass is placed from neutral axis of the bone. If both periosteal and endocortical surfaces expand in parallel, cortical thickness of the enlarging bone will remain unchanged. If endocortical expansion is excessive relative to the extent of periosteal expansion, the enlarging bone will have a progressively thinner cortex, which is not good. If periosteal expansion occurs with no endocortical expansion, or less periosteal expansion occurs with endocortical contraction, the same cortical thickness is produced. However, the latter bone will be smaller, and its cortical mass will be nearer the neutral long axis of the bone so that bending strength will be less.
Periosteal apposition increases bone width in boys and girls during
prepubertal growth. Puberty is associated with accelerated periosteal
apposition with less endocortical expansion in boys, resulting in
enlargement of the bone diameter, cortical thickening, and an increase
in the medullary diameter. Most of the increase in cortical thickness
is achieved by accelerated periosteal apposition, a process that is
likely to be regulated by T directly (via the androgen receptor), GH,
and IGF-I (17, 18). In addition, aromatization of T may
influence GH and IGF-I. When females enter puberty, periosteal
apposition is inhibited, probably due to the inhibitory effect of
estrogen on periosteal bone formation, whereas endocortical bone
formation is stimulated, increasing cortical thickness and narrowing
the medullary cavity (17, 18, 19, 20). This gender-specific and
surface-specific behavior was documented 33 yr ago in beautiful studies
of the surfaces of the metacarpal bones (Fig. 4) (19). Endocortical
contraction may also be region specific, occurring less at
weight-bearing regions such as the femur (14, 21), than
metacarpals or radius (8, 19).
|
In a recent issue of the JCEM, Schoenau et al.
(23) reported the results of a study of the growth in mass
and strength of the proximal radius in 273 females and 187 males aged
640 yr. Bone mass, measured by peripheral quantitative computed
tomography, was expressed as BMC of a 2-mm thick slice of cortical bone
measured at the distal radius. Gender differences emerged at the age of
1617 yr. BMC increased in men and women because the bone slice
widened by periosteal expansion in males and females and endocortical
contraction in females. Males had 20% higher BMC due to placement
of bone further from the neutral axis of the long bone. Bone strength
increased in both genders, but the gender difference in strength was
the result of the gender difference in bone geometry, not bone mass.
The same authors report constant volumetric BMD of the radius in an
earlier study (9), so that the literature is consistent in
this notion of the constancy of the amount of bone in the growing bone
(volumetric BMD).
Thus, the absolute and relative growth of these surfaces varies by gender, pubertal stage, the type of bone, and probably varies in degree at every position along the length of bone from proximal to distal, medial, lateral, anterior, and posterior, so that these long bones differ in size, shape, cortical thickness, and medullary diameter throughout their whole length. At each level, these traits will be determined by the interaction between bone formation on the periosteum, and the interaction of endocortical bone resorption and formation adjacent to marrow. For the endocortical expansion to occur, resorption must exceed bone formation; for endocortical contraction to occur, bone formation must exceed resorption. If resorption and formation are equal there will be no net movement of the endocortical surface.
The complex and irregular periosteal and endocortical perimeters of long bone shafts are created by the differing modeling and remodeling occurring in adjacent regions that share the same marrow. An enigmatic example of this surface-specific modeling and remodeling is described in one of the few histomorphometric studies of skeletal growth (11). Outward growth of the ilial wings of the pelvis is achieved by periosteal apposition and endocortical resorption on the outer cortex and endocortical apposition and net periosteal resorption on the inner cortex. What regulates endocortical resorption of the outer cortex and periosteal apposition on the inner cortex when these surfaces share the same marrow?!
Clinical relevance of the differing behavior of bone surfaces
during growth. The gender differences in the behavior of these
surfaces at puberty suggest that delayed puberty produces
gender-specific structural abnormalities. To speculate, boys will lose
the pubertal component of periosteal apposition, leaving a smaller bone
with a thinner cortex but normal medullary diameter (Fig. 5, top). Areal BMD will be
reduced because bone size is reduced, and volumetric BMD may be normal
if the reduction in size and accrual are proportional. Girls will lose
the pubertal endocortical apposition, leaving a normal or larger bone
(periosteal apposition continues in absence of the inhibitory effect of
estrogen) with a thinner cortex and larger medullary diameter (Fig. 5
, bottom). Thus, areal BMD may be normal or reduced depending
on bone size, but volumetric BMD will be reduced because the normal or
larger bone will have less bone mass accrual within it. Because of
reduced bone size in boys and normal (or larger) bone size in girls,
the biomechanical consequences of delayed puberty may be worse in boys
than girls, even though areal BMD may be similarly reduced.
|
Aging and the behavior of the periosteal and endosteal surfaces
During aging, periosteal apposition continues as it did during growth but much more slowly. Bone remodeling occurs at discrete sites or basic multicellular units (BMUs) on the trabecular, endocortical, and intracortical components of the endosteal (inner) surface of the skeleton. However, the remodeling within each BMU is imperfect because the volume of bone removed by bone resorption is not matched by the same volume of bone formed within the BMU. The resulting negative bone balance within each BMU is the structural basis of irreversible bone loss. The loss of bone within each BMU produces cortical thinning and intracortical porosity, especially near the bone marrow, trabecular thinning, complete loss of trabecular plates, and loss of connectivity.
The prevailing view derived using bone densitometry is that men lose
less bone than women at the spine (Fig. 6, top) (31).
This notion is produced by the fact that densitometry summates the bone
deposition on the periosteal surface plus the loss occurring inside
bone. Examination of the age-related changes that occur within the
trabecular compartment inside bone and on its periosteal surface
reveals that the amount of trabecular bone lost during aging in women
and men is similar, or only slightly less in men than women (Fig. 6
, middle) (32). This has been confirmed using
histomorphometry at the iliac crest, quantitative computed
tomography at the spine, and ashing of vertebrae
(32, 33, 34, 35, 36).
|
The periosteal bone formation occurring during aging offsets bone loss from the endosteal surfaces in both men and women, but for reasons that are not understood, periosteal apposition is greater in men than women so that the (similar) loss of bone inside bone is more greatly offset in men than in women (4, 5, 6). For example, Ruff and Hayes (4) report greater gender differences in periosteal bone formation than endosteal bone resorption at tibia. Per decade, tibial cross-sectional area (reflecting absolute periosteal bone formation) increased by 2.5% in men and 1.1% in women. Endosteal area (reflecting absolute bone resorbed) increased similarly in men and women (7% vs. 8%, respectively).
Thus, the lesser decrease in areal BMD at the spine in men than women
is due to the lesser decrease in cortical bone in men than in women
(Fig. 5, bottom) (32). However, the lesser
decrease in cortical mass is the result of greater periosteal bone
formation in men, not less resorptive removal of bone from the
endosteal surface.
The densitometer is blind to all this surface-specific bone formation and resorption. The lesser diminution in areal BMD in men than women across life in cross-sectional studies is responsible for the belief that men lose less bone than women. The missing word is "net"net bone loss is less in men than women because men form more bone during aging on the outside surface of the bone, not because the lose less bone on the inside of the bone.
The gender difference in peak bone size is not why fewer men than women have fragility fractures. The larger skeleton achieved during growth in men produces a stronger bone than in women (i.e. a bone that tolerates larger absolute loads). But, this is not the explanation for the lower incidence of fragility fractures in elderly men than in elderly women. An elephant has a bigger and stronger bone than a mouse but has to cope with the forces generated by an elephants bigger muscle. A fall on the outstretched hand loads the distal radius of a man with the kinetic energy generated by the mans height and weight, not a womans height and weight. The absolute load imposed on the vertebral body is greater in young men than women because men are taller and heavier (3, 5). However, the relative loadthe load per unit area (stress) imposed on the vertebral bodyis no different in young men and women, because the larger bone is subjected to correspondingly larger loads (5). The ratio of periosteal perimeter of the radius to muscle area is the same in boys and girls (37). There is scaling in naturea bigger bone has a bigger musclea relationship that is likely to be largely determined by genes regulating size, not necessarily the amount of exercise undertaken (38).
Fragility fractures are uncommon in young men and women because loads (determined by weight, height, and bending moments) are well below the ability of the bone to withstand them (determined by cross-sectional area and volumetric BMD). Structural failure emerges during aging in men and women because of the changing relationship between the imposed load and the bones ability to tolerate that load.
Periosteal apposition has two effects, it increases the cross-sectional area of the bone more in men, so that load imposed per unit area decreases more than in women. Second, greater periosteal apposition adds more bone to the outer perimeter of the bone in men than women offsetting the age-related endosteal bone loss more in men than women so that the fall in volumetric density of the whole bone is less in men than in women.
Thus, among men and women surviving to endure old age, both the load imposed per unit area of bone decreases more, and strength of the bone decreases less, in men than in women so that a lower proportion of elderly men than elderly women have bone size and volumetric BMD reduced below a critical level (or fracture threshold), where the loads on the bone are greater than the bones structural ability to tolerate them (5). Structural failure occurs less in men than and women because the relationship between load and bone strength is better maintained in men than in women (3, 5).
Patients with fragility fractures have reduced volumetric BMD, but they may also have reduced bone size relative to controls (39, 40, 41, 42). As the bone size difference is not taken into account in the BMC measurement, and only partly taken into account in the areal BMD measurement, the deficit in bone mass is exaggerated if the bone size is reduced relative to controls. This exagerated deficit will be attributed to reduced accrual, or (as usually the case) to excessive bone loss. Thus, the investigator is misled by the areal BMD deficit if the effect of the smaller bone size is not appreciated. If bone size in fracture cases is increased relative to controls, as reported in women with hip fractures (Duan, Y., and E. Seeman, submitted for publication), then the deficit in BMC or areal BMD will be underestimated and may obscure any deficit in volumetric BMD due to reduced accrual or excessive bone loss. Failure to account for differences in bone size in fracture cases and controls will then lead the investigator to falsely conclude that there is no deficit in volumetric BMD and, therefore, no disease process producing a deficit in volumetric BMD.
A fascinating possibility is that both deficits in bone size and volumetric BMD in women and men with spine fractures are the result of reduced periosteal apposition during aging (3). Reduced periosteal apposition during aging will 1) result in a smaller bone that tolerates bending loads less well, and will 2) fail to offset endosteal bone loss so that volumetric BMD (of the whole bone) will be lower than it otherwise would be had periosteal apposition occurred.
Summary: study the surfaces of the skeleton
Growth produces a bigger not denser skeleton in males and females
(Fig. 7). Greater periosteal growth
during puberty produces a skeleton that is just the right size to
accommodate the larger skeleton of the male. The smaller skeleton of
the female is just the right size to accommodate the smaller skeletal
mass fastened to its surface. Periosteal apposition assembles most of
the cortical mass in males. Endocortical apposition at puberty
assembles about 25% of cortical thickness at nonweight-bearing bones
in females nearer the neutral axis where it confers less biomechanical
advantage during bending than a comparable amount produced by
periosteal apposition. Less strength is needed at nonweight-bearing
sites, and it makes sense that endocortical apposition contributes
proportionally less cortex at weight-bearing sites. Whether this
estrogen-dependent endocortical accrual is the reserve for
skeletogenesis of a fetus without compromising the mothers bone
strength is a plausible hypothesis in need of testing.
|
Bone densitometry is an important tool for the clinical practice of fracture prevention. It has made valuable contributions in terms of quantitating the age- and gender-specific normal distribution of bone mass, providing a quantitative definition of "osteoporosis" and so defining the size of the problem of osteoporosis. Areal BMD is a predictor of fracture risk and so is indispensable in identifying individuals at risk of fracture who should be considered for treatment. These are valid and useful clinical applications of the method.
For clinical research, the excessive dependency on bone densitometer creates problems in understanding the structural basis of underlying the attainment of bone strength during growth, bone fragility during aging, as well as the structural basis of the slowed progression of bone fragility associated with antiresorptive drug therapy because densitometry is blind to bone size and shape, critical determinants of bone strength.
Osteoporosis begins by picking the wrong mother and father, it gains expression in the follies of a misspent youth and the excesses of adulthood, and strikes those forced to endure the futility of old age, inheriting its gifts, like senility and fractures. The word "osteoporosis" carries dangers for those interested and challenged by the need to understand the pathophysiology of bone fragility. Replace the word "osteoporosis" with "bone fragility"(... the epidemiology of ... bone fragility, the pathogenesis of ... bone fragility, the prevention and treatment of ... bone fragility); this terminology conveys the true sense, intimidating breadth, depth, and complexity of the pathophysiology of growth and aging of the skeleton.
Unless the complexity of the structural basis of attainment of bone strength during growth, and its decay during aging, is recognized and studied, young investigators entering the field will search for a single cause of a single disease where there is no one of either. The absolute and relative movement of the periosteal and endosteal surfaces produced by bone formation and bone resorption during growth and aging determine the size, mass, geometry and architecture of the skeleton, and, so, its strength. The nobility of the task will remain unfathomable while the face of bone fragility remains invisible, hidden in the vagaries of "osteoporosis" and structural ambiguities of "low areal BMD." When these terms are abandoned, the wider tapestry of bone fragility will appear resplendent in its variety.
Acknowledgments
Footnotes
Abbreviations: BMC, Bone mineral content; BMD, bone mineral density; BMU, basic multicellular unit.
Received January 18, 2001.
Accepted May 1, 2001.
References