A Structural Approach to Renal Bone Disease

Abstract

WHILE NO ONE WOULD DENY that there has been great progress made in our understanding of metabolic bone disease and our ability to evaluate these disorders in affected subjects, the accompanying editorial by Dr. Michael Parfitt provides undeniable evidence that we must recognize that progress has mixed benefits. Indeed, he points out what we have lost by reducing the use of bone biopsy and standard radiographies while relying on dual-energy X-ray absorptiometry of the central skeleton. His insights are extraordinarily relevant to the clinical decision making in patients with the bone abnormalities typically encountered in chronic renal failure. Clearly this article is essential reading for nephrologists, endocrinologists, rheumatologists, orthopedists, and bone basic scientists. Moreover, insights provided are a tribute to the gifted perspective that Dr. Parfitt brings to the bone and mineral community. Marc Drezner, M.D. Editor-in-Chief THE BONES ARE ORGANS composed of bone, which is a hard and rigid tissue. The primary function of the bones is to resist the mechanical forces applied to them by muscle contraction and gravity, so that the parts of the body can move without breaking.1 To carry out this function, each bone has a species-specific size, shape, and internal structure, the outcome of both evolutionary adaptation in the population and physiologic adaptation in the individual during growth.2,3 The precise three-dimensional location of each element of a bone is critical to its mechanical function, but much less important for subsidiary functions such as support of hematopoiesis and participation in mineral homeostasis.4 Endocrinologists and nephrologists have always been more interested in such nonmechanical functions and so have paid more attention to bone as a tissue than to bones as organs. The recent widespread recognition that increased susceptibility to fracture is the most important clinical manifestation of all metabolic bone disorders should logically have redirected attention to bones, but two aspects of current medical practice have counteracted this change in emphasis. First, most physicians no longer look at bone radiographs themselves, and so no longer observe directly the structural consequences of metabolic bone disease. Second, the ready availability of bone densitometry, with its misleading and often frankly erroneous terminology and units, has encouraged physicians to make diagnostic and therapeutic decisions on the basis of an abstract set of numbers that are completely divorced from the underlying structural reality which the numbers purport to represent. In each bone, the solid portion is enclosed between an outer periosteal envelope, which includes articular cartilage as well as periosteum, and an inner endosteal envelope, which surrounds the bone marrow and its extensions and includes the cancellous, endocortical, and intracortical subdivisions which are in continuity.5 The volume of matrix in a bone is defined by the precise three-dimensional locations of these envelopes. Since bone, unlike cartilage, is incapable of interstitial growth, primary changes in these locations underlie all changes in the size, shape, and orientation of a bone and of its internal structural elements. When bone is added, regions of bone surface usually become further apart, and when bone is lost, they usually become closer together. These changes in surface location are the result of cellular activity that takes place exclusively on surfaces, either internal or external. Bone is hard, rigid, and radiopaque because the matrix is impregnated with apatite-like mineral, which accumulates at the expense of water so that matrix volume does not change. The density, which is mass per unit volume, of unmineralized bone matrix is about 1.10 g/cm3; this increases to about 2.35 g/cm3 when all free matrix water has been replaced by mineral.6 Such complete mineralization is normally prevented by some property of living osteocytes,7 so that density does not ordinarily increase much above 2.05 g/cm3.6 For greatest accuracy, the volume to which mass is referred should exclude osteocyte lacunae and canaliculae, but in practice this refinement is usually disregarded. True bone density, as just defined, decreases if bone turnover is high and mean bone age is low, or if mineralization is impaired, and increases if bone turnover is low and mean bone age high, or if there is extensive osteocyte death.8 True bone density, the density of bone as a substance or material, must be distinguished from apparent bone density, which is the density of a whole bone as an organ, representing mass divided by external volume.8 Unlike true bone density, which changes little with age and may even increase, apparent bone density falls progressively with age because external volume either does not change or increases slightly, but matrix and mineral mass decline as bone substance is lost. Because true bone density is approximately constant in most circumstances, mineral mass, which is amenable to measurement in vivo, is a reasonable surrogate for bone matrix volume and bone mass. Apparent bone density—the bone within the bone9—is highly correlated with various measures of bone strength10 and would be an excellent index of the biologic phenomenon of age-related bone loss and of clinical fracture risk. Unfortunately, apparent bone density cannot be measured in vivo because, although mineral mass can be measured with high precision, volume can only be estimated quite crudely.11 Most measurements of so-called bone density are in fact measurements of bone mineral mass partly normalized by some index of bone size. The semantic confusion is so great that one editorialist12 referred to apparent bone density as “true” (meaning three-dimensional) bone density! It has even been suggested13 that the use of the term “bone density” be abandoned! This bizarre situation—people who claim to measure bone density not even knowing what density is—persists because the use that physicians make of numbers depends not on their names, units, or scientific meaning but on how they relate to a reference range. The structural changes that occur in bone as a result of age itself and of age-related changes in muscular and hormonal function are sufficiently uniform that they can be captured by the numbers that are generated by bone “densitometry.” Relating such numbers to bone size is crucial to the understanding of skeletal growth14 but is of much less importance in the adult skeleton. Such numbers, whether or not they are corrected for bone size, provide useful information about bone strength and future fracture risk, although not as much information as when they are related more directly to mechanical properties.15 But the underlying assumptions, that true bone density does not change and that a particular fall in numerical value has the same structural significance in different persons, may no longer hold when the bones are affected by specific diseases as well as by aging. This possibility is greatest in the most complex and least predictable form of metabolic bone disease, that resulting from chronic renal failure.16 Retardation of growth and rickets-like deformities in children and rapidly progressive genu valgum in adolescents were noted more than a century ago and remained the most frequent skeletal manifestations of renal failure for many decades thereafter.17 More recently, radiographs revealed growth plate lesions resembling rickets, looser zones of osteomalacia, subperiosteal, subchondral, and subtendinous erosions, and cystic lesions due to osteitis fibrosa and osteosclerosis, especially in the endplates of the vertebral bodies.18 All these were noted in patients with end-stage renal failure before the availability of maintenance hemodialysis, which in addition to prolonging life led initially to a specific form of osteomalacia due to aluminum poisoning.19 Recognition of this disorder and its etiology required histologic examination of bone, an investigative tool first widely applied to renal bone disease by the Manchester group.20 The same tool revealed the decline in aluminum-related bone disease that followed appropriate preventive measures21 and the emergence of so-called aplastic or adynamic bone disease, in which bone formation is reduced but mineralization is not impaired.22 A variety of noninvasive tests have been tried, alone and in combination, to predict bone histology, and in many dialysis centers these have largely replaced bone biopsy as a guide to management, a trend that has been strongly criticized.23,24 The fall in use of bone biopsy has been accompanied by a rise in use of bone “densitometry.”25-28 The earliest in vivo structural study of bone was carried out by measurement of metacarpal cortical dimensions on hand radiographs. Although imprecise, this procedure had the unique advantage of surface specificity and established the most important facts about age-related bone loss as a biologic phenomenon.29,30 The earliest densitometric method was single-energy photon absorptiometry of the radius, a site of predominantly cortical bone. The development of dual-energy photon absorptiometry, and later dual-energy X-ray absorptiometry, allowed measurement of the spine and upper femur, sites with a higher proportion of cancellous bone and where (not necessarily as a result) the most important osteoporotic fractures occur. For these reasons, in the osteoporosis field, emphasis shifted strongly away from the peripheral toward the central skeleton.31 The same shift occurred in the study of renal bone disease, not because the central skeleton was the site of the most clinically significant loss of bone mineral, but simply because it was fashionable. By the time these methods were available, most nephrologists, like most endocrinologists, had either lost or never acquired the ability to think about bone in precise structural terms and so were unable to relate the results of “densitometry” to what was actually going on in the bones. The numbers, so readily obtained, were still just numbers. The histologic study of bone can open a window on its structure, but regrettably its practitioners have unnecessarily restricted their field of vision. Except for very limited use of the rib,32 bone samples were (and are) obtained from the ilium, a site preferred because it was accessible, safe, and representative of the central skeleton.33 The information that can be obtained from an iliac biopsy depends on the location and size of the sample. When obtained through the crest by Jamshidi needle or electric drill,34 only cancellous bone is obtained. Examination of such samples confirmed the high frequency of bone changes due to hyperparathyroidism, ranging from mild increase in turnover to severe osteitis fibrosa with dissecting resorption, abundant and more highly nucleated osteoclasts, para- and intratrabecular fibrosis, and in extreme cases, woven bone formation.23,35 Small samples of cancellous bone are also satisfactory for the recognition of osteomalacia and the detection of aluminum deposition. Bone structure is better preserved in transiliac specimens with cortex at each end,33-35 but the cortical bone is generally regarded more as protection for the cancellous bone than as worthy of examination in its own right. Thus, in the current study of renal bone disease, invasive as well as noninvasive methods focus on cancellous bone to the virtual exclusion of cortical bone. Cortical bone, or compacta, forms the almost solid outer wall of all bones. Most mechanical load bearing in the skeleton is carried out by cortical bone, which bears the immediate burden of all skeletal muscle contraction, because all muscles are attached, directly or indirectly, to periosteum. Even at sites with the highest proportion of cancellous bone, such as the long bone metaphyses and vertebral bodies, fractures began in cortical bone.36 In the skeleton as a whole, cortical bone provides about 75% of the volume and mass but only about 25% of the surface,5 but in the ilium, representative of the central skeleton, the proportions are closer to two-thirds and one-third, respectively.37 The relative loss of bone with age is greater for cancellous than for cortical bone, because of its higher surface-to-volume ratio, but the absolute amount of bone lost is greater in cortical bone. This is because the volume of bone lost per unit of surface is greater for the endocortical than for the intracortical and cancellous subdivisions of the endosteal envelope.38 Furthermore, thinning of vertebral cortices with age contributes substantially to loss of compressive strength of the vertebral bodies,39 and much of the strength of the spinal column as a functional unit is provided by the neural arches and their spinous processes, which consist primarily of cortical bone. The usual reason given for the examination of cancellous rather than cortical bone is that because of its higher turnover, changes due to disease occur sooner and are of larger magnitude.33 The literature is replete with unqualified statements that bone turnover is higher in cancellous than in cortical bone, but the reality is not so simple. Bone turnover, which is fractional volume replacement per unit time (usually expressed as percentage per year), depends on two unrelated factors. These are a geometric factor-surface to volume ratio (BS/BV) and a cellular factor-bone formation rate per unit of surface (BFR/BS). BS/BV is invariably higher in cancellous than in cortical bone, although for both types there is some variability between sites. In the ilium of healthy young women, the ratio is about 4:1.37 In the same subjects, BFR/BS was about 50% higher on the intracortical and endocortical surfaces than on the cancellous surface, so that bone turnover in the ilium was only 2.5–3 times higher in cancellous bone (about 15%/year) than in cortical bone (about 5–6%/year).38 Remodeling in cancellous bone is much lower adjacent to yellow than to red marrow, so that turnover is higher in central cortical bone than in peripheral cancellous bone.40 Nevertheless, when compared at the same sites, osteoid accumulation in disease is substantially greater in cancellous than in cortical bone.41 The structure of cortical bone can be economically described by two features: thickness, which varies substantially from <1 mm to >10 mm at different skeletal sites, and porosity, which is usually in the range of 2–8% and varies much less between sites. Cortical thickness at skeletal maturity depends mainly on the magnitude of combined periosteal and endocortical net apposition during the adolescent growth spurt.3 Cortical thinning with age reflects the balance between very slow periosteal expansion and much more rapid net endocortical resorption.13,29 The major cellular mechanism is that during each cycle of remodeling, resorption cavities are deeper than at an earlier age. This has been demonstrated at both commonly used biopsy sites, ilium38 and rib,42 and there is no reason to doubt that it is a universal characteristic of age-related cortical bone loss. On bone surfaces in contact with marrow, the basic multicellular unit, which is the instrument of bone remodeling, travels across the surface excavating a trench.40,43 During slow cortical thinning, the trenches are deeper, most likely because of delayed osteoclast apoptosis,44 and are incompletely repaired. During rapid cortical thinning, as after menopause, the osteoclastic cutting hemicone tunnels under the surface, changing the direction of the trench, leading to expansion and eventual confluence of subendocortical cavities and conversion of the inner third of the original cortex to a cancellous-like structure.45,46 In the early stages of this process, cortical porosity increases substantially, but porosity in the cortical bone that remains increases only slightly with age.37 At any age a temporary increase in cortical porosity will follow any increase in bone turnover.8 Changes in cortical bone due to hyperparathyroidism, whether primary or secondary to disordered vitamin D and calcium metabolism, depend on its severity. If mild to moderate, the normal effects of age and menopause are exaggerated in a nonspecific manner. If severe, additional qualitative changes occur that are specific.47 In the former case, bone turnover is increased about 2- to 3-fold, which amplifies proportionately whatever rates of change are already established.48 There are more remodeling sites on the endocortical and intracortical surfaces because each basic multicellular unit travels further beyond its target.40,49 The trenches on the endocortical (but not intracortical) surfaces are deeper because the effects of estrogen deficiency and parathyroid hormone (PTH) excess on the timing of osteoclast apoptosis44 are additive at this site. As a result, cancellization of the inner cortex is more extensive, the rate of endocortical bone loss is considerably increased, and cortical thickness declines, despite a modest increase in net periosteal apposition.50,51 Cortical porosity is increased as in any other situation of high bone turnover.48 Patients with fortuitously discovered primary hyperparathyroidism have lost cortical bone but are no longer losing it faster than normal.52 This implies a labile compartment of inner cortical bone that is lost more quickly, in the same location as the bone that was added during the pubertal growth spurt.3,29 The qualitative changes of severe hyperparathyroidism are referred to collectively as osteitis fibrosa, an etymologically inexact but nevertheless useful term. The most distinctive feature of osteitis fibrosa in cortical bone is subperiosteal erosion, most readily detected in the phalanges. Shallow remodeling occurs infrequently beneath normal periosteum, where slight irregularity is sometimes just visible on high resolution radiographs of the hand.53 But in severe hyperparathyroidism, the cavities are much deeper and more extensive18 and histologically are filled with a mixture of fibrous tissue and loosely textured woven bone, both mineralized and unmineralized.35 Increased osteoclast penetration occurs also on all three subdivisions of the endosteal envelope, with larger intracortical cavities, even more extensive cancellization of the inner cortex, and dissecting intratrabecular resorption with undermining of the surface.47 The amount of cancellous bone is often increased because of thicker trabeculae, and corticomedullary differentiation is blurred and in extreme cases lost altogether. In mild hyperparathyroidism, in contrast to cortical bone, the age-related changes in cancellous bone are not exaggerated but rather minimized, with fewer perforations and greater preservation of connectivity.54 Thus, mild hyperparathyroidism and estrogen deficiency have similar effects on bone turnover throughout the skeleton, and similar structural effects on cortical bone, but opposite structural effects on cancellous bone.49 The major consequence of chronic renal failure with the potential for adversely affecting bone is secondary hyperparathyroidism. Increased PTH secretion begins very early in the course, when other abnormalities in bone and mineral metabolism are minimal or absent and the patients are asymptomatic.55 The pathogenesis of secondary hyperparathyroidism is complex16 and need not be discussed here. More important to emphasize is its frequency; by the time creatinine clearance has fallen below 50 ml/minute, PTH hypersecretion of some degree is almost universal.55 In such patients, the state of the bones should be the same as in mild asymptomatic primary hyperparathyroidism but has been studied infrequently. Occasional comments on cortical thinning can be found,30,56,57 but as far as I can determine, only one systematic study has been carried out in asymptomatic patients.50 Between 1962 and 1970, the author obtained hand X-rays on every patient admitted to a general medical service in Brisbane, Australia who had a stable plasma creatinine of >1.5 mg/dl but no clinical evidence of renal failure. In 86 such patients, metacarpal morphometry performed in Detroit between 1973 and 1975 showed increases in both outer and inner diameters and a significant reduction in mean combined cortical thickness of 0.78 mm or about 15%, with a reduction in relative cortical cross-sectional area of about 11%.50 PTH was not measured, but metacarpal cortical thickness correlates inversely with serum total alkaline phosphatase57 and no alternative mechanism is plausible. None of the 86 patients had subperiosteal erosion, which needs higher PTH levels than the endocortical changes. Because of this difference, the two surfaces can behave independently. For example, partial control of PTH hypersecretion by calcitriol therapy58 or renal transplantation59 may allow subperiosteal erosion to heal but cortical thinning to progress. Hand radiographs are uniquely suited to the study of renal bone disease, revealing not only subperiosteal and endocortical changes but metacarpal striation, indicating increased cortical porosity, a reflection of increased intracortical remodeling.53 By the time dialysis is needed, many patients have changes on all three surfaces of cortical bone,53,58 but the endocortical changes are the most consistent, representing the largest component of bone loss, and unlike the periosteal and intracortical changes, are irreversible.41,47 Cortical thinning is much the largest component of the low values for radial single-energy photon absorptiometry found at the onset of dialysis therapy, at both proximal and distal sites,60-62 although the magnitude of the deficits is less if conservative management is improved and if dialysis is started earlier. Further cortical thinning can be prevented by an appropriate treatment regimen, of which an adequate dialysate calcium concentration is the most important component.56,60,61 In selected patients, cortical bone loss can not only be arrested but partly reversed by parathyroidectomy.62 Unlike both aging and estrogen deficiency, whose effects on cortical and cancellous bone are directionally similar although differing in structural detail,37,45 PTH excess has generally catabolic effects on cortical bone and generally anabolic effects on cancellous bone. This contrast, already summarized for primary hyperparathyroidism, is even more striking for uremic secondary hyperparathyroidism. Cancellous osteosclerosis with thicker trabeculae has long been recognized as a feature of renal bone disease,17,18,35,59 and strong circumstantial evidence was assembled that incriminated PTH excess as a major etiologic factor.59 Osteosclerosis was more common and more severe than in primary hyperparathyroidism, partly because PTH levels are higher and partly because of the independent contribution of hyperphosphatemia to osteoblast function.59 Cortical osteopenia affecting mainly the appendicular skeleton and cancellous osteosclerosis affecting mainly the axial skeleton together represent a redistribution of skeletal assets with little change in total body calcium or total skeletal mass.59 The bone anabolic effect of PTH is now more definitely established and its mechanism better understood. PTH acts on early osteoblast progenitors to increase osteoblast recruitment16; as with sodium fluoride, there is both de novo bone formation at previously quiescent bone surfaces and over filling of resorption cavities, reflected by an increase in wall thickness.63 Why these effects are restricted to cancellous bone surfaces is unknown. Osteosclerosis features less prominently in current than in previous accounts of renal bone disease.16 It may indeed be less common, because efforts to reduce PTH secretion begin earlier and are more effective, but it is also less well recognized.64 Skeletal radiographs are now rarely ordered and even more rarely examined by nephrologists. Instead, reliance is placed on densitometric methods such as dual-energy photon absorptiometry or dual-energy X-ray absorptiometry, which in the spine lump together the mainly cancellous bone of the vertebral bodies and the mainly cortical bone of the vertebral appendages, and so conceal the most distinctive structural features. Predictably, the results have been quite variable with mean values that are higher than,28,65 the same as66,67 or, most commonly, lower than26,27,68,69 in control subjects. The most consistent finding is a wider scatter of results, with a higher than usual proportion of both abnormally high and abnormally low values.26-28,65-69 In the upper femur, the results have been generally lower than in the spine,27,28,66 especially in Ward's triangle, a predominantly cortical site,69 but with comparably wide scatter. Such studies have contributed very little, either to the understanding of pathophysiology or to the management of individual patients, but have helped in the evaluation of different treatment regimens.65-69 More informative have been studies with quantitative computed tomography, which can measure apparent density in a small cube of cancellous bone in the center of a vertebral body.69-71 The results correlate highly with bone volume in tissue volume (BV/TV) in iliac bone biopsies, including patients with excess osteoid.70 The mean value was increased (z = +1.6) in patients with histologic evidence of osteitis fibrosa, exemplifying the anabolic effect of PTH.71 The mean value was decreased (z = −1.2) in patients with low turnover osteodystrophy, which includes both generalized and atypical osteomalacia as well as aplastic bone disease.72-74 Most of the individual values were in the lower half of the normal range.71 It is likely that much of the mineral deficit reflected unmineralized osteoid, since in aplastic bone disease iliac BV/TV is normal.73,74 Apart from oversuppression of PTH, the cause of aplastic bone disease in unknown. Its clinical significance is also still unclear,21,22,73,74 although the inferred abnormality of osteoblast function might contribute to the impaired connectivity and increased vertebral deformity found at autopsy after prolonged hemodialysis.75 Osteoporotic vertebral deformities have been observed on radiographs,64 but whether their prevalence is higher in dialysis patients than in age-, gender-, and race-matched control subjects, or is related to histologic classification, remains undetermined. Early accounts of renal osteodystrophy did not mention fractures,17,20 most likely because, before renal replacement therapy became available, patients with renal failure did not live long enough for any increase in fracture risk to become manifest. Later, there were isolated reports of pathologic fractures, sometimes precipitated by seizures, through focal cystic lesions of osteitis fibrosa.76,77 More recently, hip fractures have occurred through regions of amyloid infiltration, although only after dialysis for at least 5 years.78,79 But fractures in significant numbers were first reported in association with aluminum-related osteomalacia,19,59,80,81 ribs being the most common site.82 The biomechanics of these fractures were not studied, but complete fractures through Looser zones have long been known to occur in osteomalacia,83 and replacement of a substantial amount of mineralized bone by osteoid would be expected to significantly reduce bone strength.84 Nevertheless, when all tissue in an iliac biopsy core is accounted for, cortical thinning makes much the largest contribution to total mineralized bone loss, even in patients with severe aluminum-related osteomalacia.85 Several series of hip fractures in dialysis patients have been reported,79,86 but there has been only one comprehensive epidemiologic study. Based on hospital discharge codes, hip fracture risk is increased 3- to 4-fold in patients with end-stage renal disease, regardless of gender or race.87 The relative risk was highest (6- to 14-fold) in the 30- to 39-year-old age group, but the absolute or attributable risk was highest in the 70- to 84-year-old age group. Individual fracture risk may be related to the severity of amyloid disease78 or to vitamin K status and apolipoprotein polymorphism,88 but such a large increase in hip fracture risk would require some factor affecting the dialysis population as a whole. Based on the preceding discussion, the most likely such factor is generalized cortical thinning, which is already severe before the onset of dialysis therapy, and will remain the major determinant of bone strength regardless of subsequent developments and regardless of the histologic classification or the disorder of bone remodeling that may be present at the time the fracture occurs.85 Although oversuppression of PTH secretion may be harmful during dialysis therapy,16,73 judicious parathyroidectomy is beneficial,62 and prevention of PTH hypersecretion much earlier in the disease course offers the best hope of reducing fracture risk in patients with chronic renal failure.85