Foldes et al.(1) and Kimmel et al.(2) reported that patients with spine fractures had reduced iliac crest biopsy core width (the distance between periosteal surfaces) and reduced cortical thickness, but normal medullary width. We, not these authors, inferred that these observations may be explained by reduced periosteal apposition during aging.(3) I apologize for the unintentional attribution. I accept that Dr. Parfitt's calculations, based on bone formation rates at the iliac crest,(4) favor a growth-related origin of the smaller iliac crest cortical thickness.
I am an advocate of the growth-related origins of the structural abnormalities found in patients with fractures,(5–8) but I am reluctant to discard the possibility that reduced periosteal apposition during advancing age contributes to reduced bone size for several reasons. First, sex differences in periosteal apposition rather than endosteal bone loss account for a greater amount of the sex difference in net bone loss during aging.(3)Net bone loss from the vertebral body across age was less in men than women (0.6 g vs. 1.9 g). However, men resorbed more bone than women, not less, in absolute terms (3.7 g vs. 3.1 g), whereas the absolute periosteal bone deposited was greater in men than women (3.1 g vs. 1.2 g).
Second, estimated vertebral body volumetric bone mineral density (vBMD) in the elderly controls, assuming no periosteal apposition occurred during aging, was no different to the observed vBMD in spine fracture patients. This estimate lead us to “speculate that reduced periosteal bone formation may be responsible for both reduced bone size and, in part, for the reduced vBMD in men and women with spine fractures.”(3) Third, daughters of women with spine fractures do not to have reduced bone size.(6) Although this could be a type 2 error, a deficit would be expected of half the magnitude found in their mothers relative to their peers if this was a growth-related process.
Fourth, even if the contribution of reduced periosteal apposition is small, it is likely to be important in biomechanical terms. A small difference in periosteal apposition is likely to have more deleterious effects on the compressive and bending strength of bone than a greater amount of bone deposited (during growth or aging) than retained (during aging) on the endosteal surface.(9) Fifth, periosteal apposition is sex-specific and is likely to be region-specific. If age-related periosteal apposition is partly an adaptive response to increased strains caused by loading a bone diminished in mass by bone loss, then the adaptive response may vary from region to region and may be greater at the vertebra than the iliac crest, greater in some vertebrae than others, and greater in some parts of the same vertebra than others.(10)
The nature of “reduced” periosteal apposition during growth and its role in bone fragility in old age is just as neglected as is periosteal apposition during aging. Shorter individuals have narrower bones; therefore, periosteal apposition must be “reduced” compared with taller individuals, but the length/width ratio may not differ. Periosteal apposition “reduced” relative to growth in bone length may produce a slender bone compromising the ability of that bone to tolerate loads. We know little, if anything, about the coregulation of bone length, width, and cortical thickness and whether slenderness is more common in patients with fractures, and if so, what factors produce disproportionate growth in the dimensions of bone leading to slenderness. If dimensions are smaller but proportionally so, why should the stresses be greater on a smaller than larger bone if there is scaling in nature, so that each bone has a proportionally matched skeletal mass producing similar maximum stresses (load per unit area) at equivalent levels of activity.(11)
Notions of “excessive” bone resorption in fracture cases compared with controls are also on pretty shaky grounds. Medullary diameter was not increased in the fracture cases,(1,2) suggesting that the thinner cortices were not due to greater endocortical resorption than in controls.(3) Dr. Parfitt shrewdly speculated (in a personal communication, not this letter) that finding no increase in medullary diameter relative to controls may be the result of excessive bone resorption if the peak medullary diameter was smaller in the smaller bone. We examined this and found an inverse association between femoral neck width and cortical thickness (age-, height-, and weight-adjusted) (Y. Duan and E. Seeman, unpublished data, 2001); the narrower the bone, the thicker the cortex, consistent with Dr. Parfitt's speculation. But how compelling is the evidence that bone resorption is greater in patients with fractures than in controls?
An imbalance in the basic multicellular unit (BMU) produces bone loss, cortical thinning, porosity, trabecular thinning, loss of connectivity, and bone fragility. This is not controversial. However, the notion of “greater” resorption requires evidence of a more negative bone balance at the BMU due to a lower volume of bone formed in the BMU in cases than in controls, a greater volume of bone resorbed in the BMU in cases than in controls, and a higher activation frequency in cases than in controls. Histomorphometric and biochemical evidence for these differences is often lacking or weak.(12–18) Although some of these studies report a higher group mean for indices of resorption or a lower group mean for indices of bone formation in cases than in controls, the scatter of the data is more impressive than the difference in the means, suggesting heterogeneity in the pathogenesis of bone fragility.
The possibility that reduced periosteal apposition during aging contributes to reduced bone size, net bone loss, and therefore bone fragility should not be discarded. Even if endosteal resorption is greater than in controls, as discussed above, from a biomechanical point of view, endocortical bone loss is less important than failed periosteal apposition. I do not suggest that the smaller bone size is due to reduced periosteal apposition during aging, only that our observations generate “… the need to test the hypothesis…. ”(3)
Maybe we are looking in the wrong place. The solution to the problem of the pathogenesis of fracture as opposed to pathogenesis of osteoporosis resides in the structural and material properties that influence bone biomechanics: the neglected osteocyte(19); the intervertebral disc, a cushion in search of younger feathers(20); and, I believe, the periosteal osteoblast. I will discard this beautiful hypothesis when data are published reporting that periosteal apposition at the vertebral body is no different in patients with spine fractures than that observed in controls. I thank Dr. Parfitt for his insights, which are always educational and thoroughly challenging.