From Wolff's law to the Utah paradigm: Insights about bone physiology and its clinical applications


  • Harold M. Frost

    Corresponding author
    1. Department of Orthopaedic Surgery, Southern Colorado Clinic, Pueblo, Colorado 81004
    • Department of Orthopaedic Surgery, Southern Colorado Clinic, Pueblo, CO 81004
    Search for more papers by this author


Efforts to understand our anatomy and physiology can involve four often overlapping phases. We study what occurs, then how, then ask why, and then seek clinical applications. In that regard, in 1960 views, bone's effector cells (osteoblasts and osteoclasts) worked chiefly to maintain homeostasis under the control of nonmechanical agents, and that physiology had little to do with anatomy, biomechanics, tissue-level things, muscle, and other clinical applications. But it seems later-discovered tissue-level mechanisms and functions (including biomechanical ones, plus muscle) are the true key players in bone physiology, and homeostasis ranks below the mechanical functions. Adding that information to earlier views led to the Utah paradigm of skeletal physiology that combines varied anatomical, clinical, pathological, and basic science evidence and ideas. While it explains in a general way how strong muscles make strong bones and chronically weak muscles make weak ones, and while many anatomists know about the physiology that fact depends on, poor interdisciplinary communication left people in many other specialties unaware of it and its applications. Those applications concern 1.) healing of fractures, osteotomies, and arthrodeses; 2.) criteria that distinguish mechanically competent from incompetent bones; 3.) design criteria that should let load-bearing implants endure; 4.) how to increase bone strength during growth, and how to maintain it afterwards on earth and in microgravity situations in space; 5.) how and why healthy women only lose bone next to marrow during menopause; 6.) why normal bone functions can cause osteopenias; 7.) why whole-bone strength and bone health are different matters; 8.) why falls can cause metaphyseal and diaphyseal fractures of the radius in children, but mainly metaphyseal fractures of that bone in aged adults; 9.) which methods could best evaluate whole-bone strength, “osteopenias” and “osteoporoses”; 10.) and why most “osteoporoses” should not have bone-genetic causes and some could have extraosseous genetic causes. Clinical specialties that currently require this information include orthopaedics, endocrinology, radiology, rheumatology, pediatrics, neurology, nutrition, dentistry, and physical, space and sports medicine. Basic science specialties include absorptiometry, anatomy, anthropology, biochemistry, biomechanics, biophysics, genetics, histology, pathology, pharmacology, and cell and molecular biology. This article reviews our present general understanding of this new bone physiology and some of its clinical applications and implications. It must leave to other times, places, and people the resolution of questions about that new physiology, and to understand the many devils that should lie in its details. (Thompson D'Arcy, 1917). Anat Rec 262:398–419, 2001. © 2001 Wiley-Liss, Inc.

“…between muscle and bone there can be no change in the one but it is correlated with changes in the other…”

(Thompson, 1917)

We try to understand the anatomy and physiology of our body's organ systems to achieve better management of their clinical problems. In that vein, for over 100 years anatomists and orthopaedic surgeons looked to Wolff's Law (Wolff, 1892). One translation of it from German to English reads thus (Rasche and Burke, 1962): “Every change in the form and function of bone or of their function alone is followed by certain definite changes in their internal architecture, and equally definite alteration in their external conformation, in accordance with mathematical laws.” Hindsight reveals some limitations of that Law. For example, in 1892 it had no clinical applications; it said mechanical influences can affect bone architecture, but not how; and it could not predict particular effects of specific mechanical challenges.

Later evidence began to resolve such limitations and led to the still-evolving Utah paradigm of skeletal physiology that sprang from a soil of multidisciplinary evidence and ideas (Burr and Martin, 1992; Frost, 1992, 2000a; Jee and Frost, 1992; Schönau, 1996; Takahashi, 1995, 1999). It adds tissue-level and anatomical features and roles to former views that emphasized cell-level features and roles. The new paradigm has growing applications to joints, ligaments, tendons, and fascia (Frost, 1995), but this article concerns its applications to bone as material and to bones as organs. Table 1 lists some of the things the paradigm can explain. Three of its propositions concern the purpose of load-bearing bones and how that is achieved.

Table 1. Clinical phenomena the Utah paradigm can explain plausibly*
  • *

    While plausible need not mean correct too, as the number of things a paradigm can explain increases, so do its credibility and usefulness.

Why strong muscles usually associate with strong bones.
Why too stiff and too compliant internal fixations can each impair the healing of fractures, bone grafts, and arthrodeses.
What makes fracture callus reshape itself.
Why some screws or external fixation pins in bone loosen or pull out.
Why the bone supporting some endoprosthetic designs collapses.
Why most aging adults lose bone strength and “mass.”
Why chronic muscle weakness associates with an osteopenia.
Why chronic debilitating illness usually associates with an osteopenia.
Why whole-bone strength and stiffness usually correlate well.
What makes healthy bones stronger than needed for their usual voluntary mechanical usage.
What makes net bone losses usually come from bone next to marrow.
Why obese people have more bone strength than equally active slender people.
Why some patients with “osteoporosis” develop spontaneous fractures but others do not.
What causes stress fractures, “spontaneous fractures,” and pseudofractures.
Why most men have more bone strength than most women.
The chief cause of increased bone fragility in osteogenesis imperfecta.
Why “vigorous” exercise can help to keep but not increase bone strength in adults.
How to distinguish mechanically competent bones from incompetent ones.
Why osteoclasts are not essential for effective homeostasis.
Why remodeling could not cure an osteopenia but modeling could.
The direct cause of most so-called “osteoporosis fractures.”

Proposition 1:Healthy, postnatal load-bearing bones are designed to have only enough strength to keep chronically subnormal, normal or supranormalvoluntaryloads (not injuries) from causingspontaneousfractures (Frost, 1997a). Achieving that “mechanical competence” should be the ultimate test of a bone's health and the main goal of its biologic mechanisms. In that view, bone “health” and “strength” differ. The former would depend on the relationship between a bone's strength and the size of the peak loads it usually carries. Thus mouse and horse femurs that satisfied Proposition 1 in the animals they came from would be equally healthy, even though their strengths differ more than 1,000 times. Likewise for the ribs and femurs in a single human being.

Proposition 2 (in three parts): 1.) To achieve mechanical competence bone's tissue-level biologic mechanisms need nonmechanical factors and effector cells (osteoblasts and osteoclasts), just as cars need motors, fuel and wheels in order to move. 2.) In a negative feedback arrangement bone loads and strains guide those biologic mechanisms in time and anatomical space**, just as steering, brakes, and accelerators guide cars. 3.) Most nonmechanical factors can help or modulate that guidance but cannot replace it**. In proof, such factors cannot normalize bones, joints or tendons in paralyzed limbs. Equally, motors and fuel do not tell cars where to go or when.

To explain, the newborn skeleton already has its basic architecture and relationships, and the biologic mechanisms, responses, and signaling mechanisms that can adapt it to postnatal influences (Cruess, 1982; Enlow, 1963; Hanes and Mohidden, 1965; Weinmann and Sicher, 1955). The signaling mechanisms could include osteocytes, bone lining cells, and other cells in the marrow, and could depend on streaming and piezoelectric potentials, some chemical phenomena, and fluid shear over cell membranes (Marotti et al., 1996; Martin, 2000). Chiefly gene expression patterns in utero would predetermine those baseline conditions (mechanical effects in the last trimester of pregnancy are ignored here; Carter and Wong, 1990). At any time after birth, the skeletal organs in neonatally paralyzed and normal limbs show typical differences in their strength, architecture, and other features. Those differences should reveal the kinds and magnitudes of the adaptations to postnatal mechanical loads in the normal limbs. Structures in the totally paralyzed limbs should reveal the baseline conditions, presumably affected by postnatal nonmechanical agents including genes (Jaffurs and Evans, 1998), but not by normal postnatal loads (Frost, 1995).

Implication: Bones, fascia, ligaments, and tendons should not completely disappear in total permanent disuse. Indeed, in the lower limbs of patients with congenital complete paralysis due to myelomeningocele, some bone and other structural tissues always remain (Frost, 1986).

The following sections demonstrate how those postnatal bone differences develop and some clinical applications of that physiology. A Glossary at the end defines some of the terms used in the text. Note: Throughout the text, a double asterisk (**) after a statement will mean: “While some might consider that statement controversial and I will respect such views, I am sure the statement is valid.”


Four Physical Features Combine to Determine Whole-Bone Strength

(Currey,1984Martin et al,1998)

1.) That strength depends on bone's stiffness, ultimate strength, resilience, true density, etc. (the materials property factor). 2.) It depends on the kinds of bone— woven, plexiform and lamellar, compacta, and spongiosa— and their amounts in a cross-section (the “mass” factor). That amount usually increases during growth, plateaus in young adults, and then declines, so by 80 years of age less than 60% of the young-adult bone “mass” (and strength) can remain (Buckwalter et al., 1993; Marcus et al., 1996; Smith and Gilligan, 1989). 3.) Its size, shape, and the distribution of its bony tissue in space affect a bone's strength (the architectural factor). Thus, doubling outside bone diameter while keeping the same amount of bone in the cross-section so the cortex becomes thinner, increases bending strength about eight times. 4.) Fatigue damage or microdamage also affects a bone's strength (the microdamage factor) (Burr et al., 1997; Forwood and Parker, 1989; Martin 1995, 2000).

Longitudinal bone growth and the baseline conditions excepted, normally a bone's anatomy depends on “2” and “3,” and whole-bone strength depends chiefly on “2,3,4.” Why not on “1” too? When compared to the big differences in strength of our ribs and femurs, bone's materials properties change relatively little with aging, sex, species, bones, and most diseases, osteomalacia excepted (Martin et al., 1998). Like bone “mass,” normally whole-bone strength increases during growth; it plateaus in young adults and declines afterwards (Ferretti, 1999; Garn, 1970; Schiessl et al., 1998). This article emphasizes whole-bone strength, since nature seems to rank its importance above bone “mass.”

Two Biologic Determinants of Whole-Bone Strength: Bone Modeling and Remodeling

In former views, independently working osteoblasts controlled gains of bone, and independently working osteoclasts controlled bone losses (Aegerter and Kirkpatrick, 1975; Albright and Reifenstein, 1948; McLean and Urist, 1961; Snapper, 1957). But, except longitudinal bone growth, tissue-level modeling and remodeling mechanisms chiefly control those gains and losses and help to control a bone's gross anatomy. Both mechanisms need osteoblasts and osteoclasts as well as precursor, stem, and other cells to do their work (Jee, 1989; Martin et al., 1998; Parfitt et al., 1996; Schönau, 1996).

Modeling by resorption and formation drifts (Fig. 1) can move bone surfaces in tissue space to determine the cross-sectional size and shape and longitudinal shape of bones and trabeculae (Jee, 1989). That seems to be nature's preferred way to increase a bone's strength. Modeling would seldom if ever reduce a bone's strength.

Figure 1.

Bone modeling by drifts. A: An infant's long bone with its original size and shape in solid line. To keep its shape as it grows in length and diameter, drifts move its surfaces in tissue space as the dashed lines suggest. Formation drifts make and control new osteoblasts to build some surfaces up. Resorption drifts make and control new osteoclasts to remove bone from other surfaces. B: A different drift pattern can correct the fracture malunion in a child shown in solid line. The cross-section view (right) shows the endocortical as well as the periosteal drifts that do that. C: How the drifts in B would move the whole segment to the right. Changing the anatomy in that way reduces the bone's bending moments. Drifts are created when and where they are needed, and include capillaries, precursor and “supporting” cells, and some wandering cells. They are multicellular entities in the same sense as renal nephrons Reproduced from Frost (1997d) with permission of the publisher.

Remodeling by BMUs (Basic Multicellular Units, Fig. 2) turns bone over in small packets in which osteoclasts resorb some bone and then osteoblasts fill the resulting hole or excavation with new bone (Jee, 1989). This remodeling can work in at least two modes. In a “conservation-mode,” completed BMUs resorb and make nearly equal amounts of bone so no significant bone gain or loss ensues; but in a “disuse-mode,” BMUs make less bone than they resorb, but only for bone next to or close to marrow (trabecular and endocortical bone) in both children and adults (Frost, 1998b)**. This disuse-mode remodeling should cause all adult-acquired osteopenias on earth and in astronauts in orbit (Frost, unpublished data1)**. It should help to explain why, in healthy human subjects, bone “mass” can decrease over 40% between 25 and 75 years of age (Marcus et al., 1996), and over 90% of that bone loss comes from bone next to marrow. During that age span, intracortical porosity increases from ∼ 3.5% to ∼ 7% (Frost, 1969). BMUs seldom, if ever, increase bone strength and “mass.” Here one should distinguish permanent bone losses caused by disuse-mode remodeling, from temporary losses from the increased remodeling space (Heaney, 1994) that always accompanies increased remodeling-dependent bone turnover (Jaworski, 1984).

Figure 2.

Bone remodeling BMUs. A–C: An activation event on a bone surface at A causes a packet of bone resorption at B, and then replacement of the resorbed bone by osteoblasts at C on the right. The BMU makes and controls the new osteoclasts and osteoblasts that do this. D–F: This emphasizes the amounts of bone resorbed (E) and formed (F) by completed BMUs. G–I: In these “BMU graphs” (after Frost), G shows a small excess of formation over resorption. H: Equalized resorption and formation as on haversian surfaces and in “conservation-mode” remodeling. I: Net deficit of formation, as in “disuse-mode” remodeling of endocortical and trabecular bone. Bottom: These “stair graphs” (after P.J. Meunier) show the effects of a series of BMUs of the kind immediately above on the local bone “bank.” BMUs are created when and where they are needed, and include a capillary, precursor and “supporting” cells, and some wandering cells. They are multicellular entities in the same sense as renal nephrons Reproduced from Frost (1997d) with permission of the publisher.

Mechanical Control of Bone Modeling and Remodeling

(Burr,1998Burr et al.,1995Forwood and Turner,1995Frost,1990a, bJee and Frost,1992Martin et al.,1998Martin,2000Umemura et al.,1997)

Mechanical loads on bones deform or strain them, and larger loads cause bigger strains. Where strains exceed a modeling threshold range, modeling slowly increases bone strength to reduce later strains towards that range; otherwise mechanically controlled modeling turns off. Those responses make bones strong enough to keep “typical peak strains” (see Glossary) from exceeding that threshold**. Since the threshold lies below bone's ultimate strength, those responses make healthy bones stronger than needed for their peak voluntary loads. In young adult mammals, that “strength-safety factor” (see Glossary) ≈ 6 when expressed in stress terms. When strains stay below a lower remodeling threshold range, disuse-mode remodeling permanently removes bone, but only next to or close to marrow**. That causes a “disuse-pattern osteopenia” characterized by less spongiosa, an enlarged marrow cavity, and a thinned cortex, but not a decreased outside bone diameter. When strains exceed that threshold, conservation-mode remodeling begins to reduce or stop those bone losses. That prevents an osteopenia or progression of an existing one.

In such ways, strain indirectly but strongly influences the postnatal strength and architecture of load-bearing bones.

Those threshold ranges make the largest strains control modeling and remodeling effects on whole-bone strength, and make lesser strains have little effect on it (Lanyon, 1996; Martin et al., 1998; Rubin and McLeod, 1994; Torrance et al., 1994). The thresholds also provide natural criteria that help to distinguish “normal” from too little or too much bone strength**. Their existence and values would reside as genetically determined internal standards in some skeletal cells (we do not yet know which cells). During mechanical usage, strain-dependent signals from bone would be compared to those standards, and if that reveals an error a corresponding “error signal” would arise that made modeling or remodeling correct the error. How aging affects these thresholds is uncertain but under study (Raab-Culen et al., 1996). The signaling mechanisms, pathways, and cells that help to control those things now form separate fields of study in skeletal science (El Haj, 1990; Fukada and Yasuda, 1957; Marotti et al., 1996; Martin, 1995, 2000; Martin et al., 1998; Skerry, 1997).

Role of Momentary Muscle Strength in Whole-Bone Strength

(Frost and Schönau,2000Kannus et al.,1996Pauwels,1986Rittweger et al.,1999Schiessl et al.,1998Schiessl and Willnecker,1999Schönau et al.,1998)

Muscles work against such bad lever arms that it takes well over 2 kg of muscle force on bones to move each kilogram of body weight around on earth (Crowninshield et al., 1978; English and Kilvington, 1979; Lu et al., 1997; Martin et al., 1998). Ergo, the largest voluntary bone loads and bone strains come from muscles, not body weight as formerly thought (Koch, 1917). Since those strains help to control modeling and remodeling effects on bone strength, momentary muscle strength (see the Glossary) indirectly but strongly affects the strength of load-bearing bones. Or: muscle forces → bones → strains → control of modeling and remodeling. Like bone strength, usually muscle strength also increases during growth, plateaus in young adults, and then declines (Burr, 1997; Faulkner et al., 1990; Larsson et al., 1979).

That should help to explain why strong muscles usually do make strong bones, and chronically weak muscles usually do make weak bones** (Doyle et al., 1970; Frost and Schönau, 2000; Jee, 1999, 2000; Jee and Li, 1990; Jee et al., 1991; Jee and Frost, 1992; Li et al., 1990; Li and Jee, 1991; Snow-Harter et al., 1990; Yao et al., 2000). For example, most women have weaker muscles than most men, so they should have less bone strength (and “mass”) too. They do, even if gender has additional effects. As Burr (1997) and Schönau et al. (1998) noted, neuromuscular influences on bone strength were long misunderstood and minimized, but they are becoming another field of study in skeletal science. That realization led Dr. GP Lyritis in Greece to form the new International Society for Musculoskeletal and Neuronal Interactions.

Variations in how different individuals use different parts of their bony skeletons mechanically can cause variable differences in the strength and tissue dynamics of different bones (Podenphant and Engel, 1987). That helps to explain why some bones need not predict the strength of some other bones very well. Such problems puzzled many osteoporosis authorities who, because of the lingering views mentioned in the Summary (Cohn et al., 1984), sought exclusively nonmechanical explanations.

Nonmechanical Control of Bone Modeling and Remodeling

In former views, factors like those in Table 2 dominated control of the postnatal strength of load-bearing bones (Bilezikian et al., 1996; Canalis, 1993; Duncan and Turner, 1995; Favus, 1999; Huffer, 1988; Parfitt, 1993, 1995; Parfitt et al., 1996). However the omissions of such views make them suspect.

Table 2. Examples of nonmechanical factors that could influence bone adaptations to mechanical usage and strains (so they could influence bone strength and “mass” too)
  1. Medications and Other Artificial Agents

HormonesVitaminsD metabolites
Dietary calciumOther mineralsCytokines
Paracrine effectsAutocrine effectsCell-cell interactions
Amino acidsLipidsThe genome
Gene expressionEthnic originOccupation
GenderSome diseasesMalnutrition

In fact, most such factors can help or modulate but cannot replace the mechanical control of postnatal bone modeling and remodeling**. As examples, by direct actions on bone cells things like hormones, calcium, vitamin D, and genes might determine 3% to as much as 10% of a bone's postnatal strength, but mechanical usage effects on modeling and remodeling determine over 40% of it**. In proof, years after a paraplegia bones in lower but not upper extremities can lose over 40% of their original bone “mass” (Kiratli, 1996). Similar events occur after total lower extremity paralysis from anterior poliomyelitis (Frost, unpublished results), while lower limb bones of patients paralyzed by a myelomeningocele show even larger deficits (Frost, unpublished observations).

Figure 3 indicates some combined effects of modeling, remodeling, and their strain thresholds on a bone's strength.

Figure 3.

Combined modeling and remodeling effects on bone strength and “mass.” The horizontal line at the bottom suggests typical peak bone strains from zero on the left, to the fracture strain on the right (Fx), plus the locations of the remodeling, modeling, and microdamage thresholds (MESr, MESm, MESp, respectively). The horizontal axis represents no net gains or losses of bone strength. The lower dotted line curve suggests how remodeling would remove bone where strains stay below the MESr range, but otherwise would tend to keep existing bone and its strength. The upper dashed line curve suggests how modeling drifts would begin to increase bone strength where strains enter or exceed the MESm range. The dashed outlines suggest the combined modeling and remodeling effects on a bone's strength. D.H. Carter originally suggested such a curve (Carter, 1984). At and beyond the MESp range, woven bone formation usually replaces lamellar bone formation. Fx = the fracture strain range centered near 25,000 microstrain. At the top, DW = disuse window; AW = adapted window as in normally adapted young adults; MOW = mild overload window as in healthy-growing mammals; POW = pathologic overload window (Frost, 1992a). In the nearly flat region between the MESr and MESm, bone strength and “mass” change little as typical strains change. Reproduced from Frost (1997d) with permission of the publisher.

Putative Marrow Mediator Mechanism

A still-enigmatic mechanism in marrow should help to control modeling and remodeling of bone next to or close to it, but not of intracortical (haversian) and subperiosteal bone (Chow et al., 1993; Erben, 1996; Frost, 1998b)**. This could explain why endocortical bone losses expand the cross-section area of the marrow cavity in human ribs by more than 50% between 20 and 75 years of age. During normal and supranormal mechanical usage, as well as under the influence of estrogen (Fig. 4), this mediator mechanism would make conservation-mode remodeling keep existing bone and thus prevent an osteopenia or progression of an existing one. In acute disuse, or in acute loss of estrogen (Wronski et al., 1993) or androgen (Christiansen et al., 1981; Erben et al., 2000, Yao et al., 2000), or during treatment with adrenalcortical steroid analogs like Prednisone, this mechanism would help to make disuse-mode remodeling cause a disuse-pattern osteopenia**. That should explain the usual loss of bone next to marrow in women going through menopause. Figure 4 shows how those responses to estrogen and muscle can affect bone “mass” and, by implication, whole-bone strength.

Figure 4.

A bone-muscle mass comparison. H. Schiessl constructed this graph from an Argentine study of 345 healthy boys and 443 healthy girls between 2 and 20 years of age (Zanchetta et al., 1995). It plots the grams of total body bone mineral content (TBMC, an indicator of whole-bone strength) on the vertical axis that correspond to the grams of lean body mass (LBM, an indicator of muscle strength) on the horizontal axis, as determined by a Norland DEXA machine. Crosses: girls. Open circles: boys. Each data point stands for an age one year older than the data point on its left, and it shows the means for all subjects in that one-year age group. Around 11 years of age TBMC began increasing faster than before in girls. By ≈ 15 years of age, their TBMC and LBM both plateaued. Since both indices were still increasing in 20-year-old males, they ended up with more muscle and bone than the 20-year-old girls. This evidence supports the roles of muscle and estrogen discussed in the main text. It has been suggested that the extra bone stored during a woman's fertile years could serve needs of lactation more than to increase whole-bone strength. Reproduced from Schiessl et al. (1998) with permission of the publisher.

Microdamage (MDx)

(Burr and Stafford,1990Burr et al.,1997Kimmel,1993Mori and Burr,1993Pattin et al.,1996)

Repeated strains cause microscopic fatigue damage (MDx) that weakens bones. Normally remodeling BMUs replace the damaged bone with new bone, and strains below an operational MDx threshold range cause so little MDx that remodeling can repair it. When larger strains cause too much to repair, the resulting accumulated MDx causes or helps to cause all “spontaneous” and stress fractures (so “spontaneous” fractures are not really spontaneous) (Devas, 1975; Frost, 1989a; Markey, 1987), as well as pseudofractures in osteomalacia and pathologic fractures**. Such accumulations can also allow pull-outs or loosening of pedicle and other screws, or make a bone weak enough to let a minor incident (low energy trauma; Freeman et al., 1974; Greenspan et al., 1994) fracture it.

Apparently this threshold lies above bone's modeling threshold but below its ultimate strength. That arrangement would minimize MDx, and it has been argued that bone design does minimize fatigue failures (Alexander, 1984; Frost, 2000b). Controversial when first described (Frost, 1960), MDx in bone now forms another field of study in skeletal science (Fazzalari et al., 1998; Martin, 1995, 2000; Schaffler et al., 1995; Verborgt et al., 2000).

Regional Acceleratory Phenomenon (RAP)

(Frost,1983, 1986, 1995Kelly,1990Kozin,1993Martin,1987Martin et al.,1998Shih and Norrdin,1985)

This ubiquitous phenomenon is a necessary factor in the normal healing of all hard and soft tissues. Injuries and other noxious stimuli usually increase all ongoing biologic activities in the affected body region. The increases include local perfusion, cell metabolism and turnover, and any ongoing growth (Ring and Ward, 1958), modeling, remodeling, healing, maintenance, and inflammatory activities. In combination, those things comprise the RAP**.

A RAP can last from a week when caused by a small pimple to over 2 years when caused by a complex large-bone fracture or a spinal fusion. Presumably it causes the long bone overgrowth that occurs after some fractures in children (Blount, 1955; Cozen, 1990; Frost, 1997b). Failure to develop a RAP can retard healing of all tissues. Inadequate regional blood supply can cause that, but so can sensory denervation after major peripheral nerve transections or in the lower extremities of patients with diabetic neuropathy (Frost, 1986). Interestingly, motor denervation alone, as in post-polio states, does not impair development of an RAP (Frost, unpublished observations). A, RAP usually responds to local need (Hernandez et al., 1995). It causes three of the classical signs of inflammation: Edema, erythema, and increased warmth. Pathological RAPs known as algodystrophies or migratory osteoporoses also occur (Duncan et al., 1973; Langloh et al., 1973; Mailis et al., 1992; Schiano et al., 1976). They usually respond well to prostaglandin inhibitors but poorly to physical therapy (Frost, unpublished data)**.

Mechanostat Hypothesis

For over 75 million years (Romer, 1966) it seems all load-bearing bones satisfied Propositions 1 and 2 in all healthy amphibians, birds, mammals, and reptiles of any size, age, and sex**. Whatever orchestrates such a universal effect was called the mechanostat (Frost, 1987b; Jee, 2000; Martin et al., 1998). As currently viewed it would combine some of the above-mentioned features to form a negative feedback system that makes modeling, remodeling, and their thresholds increase bone strength where necessary, or remove bone when it is not needed mechanically (Frost, 1996)**. The marrow mediator, estrogen, growth hormone, androgens, drugs, and other factors might modulate how the mechanostat affects whole-bone strength, in part by modulating the above thresholds or internal standards (Burr and Martin, 1992; Slemenda et al., 1994). A car can provide a useful analogy. Its steering, brakes, accelerator, and ignition switch would be like the features mentioned above; its wheels would be like effector cells. and its fuel and engine would be like the nonmechanical things in Table 2. Its driver would be like voluntary mechanical usage.


Just as studying only its wheels could not explain why a car drove to Berlin instead of Paris, studying only bone's effector cells could seldom explain the cause of an osteopenia, osteoporosis, impaired bone healing or other bone disorder**. Aided by the modeling and remodeling thresholds, this mechanostat could tell exactly where and when a bone or trabecula needs more strength or has too much, and then make modeling or remodeling correct the local error. No hormone, other humoral agent or gene can do such things (Ferretti, personal communication, 1999). The strain range between the modeling and remodeling thresholds in Figure 3 would provide a natural definition of “normal” whole-bone strength relative to the size of a bone's peak voluntary loads. The strength and architecture of some weakly-loaded cranial bones may depend more on their baseline conditions than on the mechanostat. They include the turbinates, nasal bones, ethmoids, wing of the sphenoid, frontal and parietal bones, and inner ear ossicles.


Muscle strength and anatomy combined with neuromuscular physiology determine the size and orientation of the voluntary muscle forces on bones. Because of that and the physiology summarized above, voluntary neuromuscular activities strongly influence and could even dominate control of the major fraction of the postnatal strength of our load-bearing bones**. That should help to make bones satisfy Proposition 1. The physiology supporting those sentences could represent a kind of “quantum jump” in our understanding of Wolff's Law when it is compared to earlier views, including some of my own (Brand and Claes, 1989; Chamay and Tschantz, 1972; Evans, 1957; Frost, 1964; Jansson, 1920; Muller, 1926; Roesler, 1987; Roux, 1895; Treharne, 1981; Vico et al., 1987; Welten et al., 1994; Whedon, 1984; Woo et al., 1981; Wunder et al., 1979).

The above summary concerns some of the “what, how, and why” of our 300+-year effort to understand bone anatomy and physiology. That makes this question pertinent: Does that understanding have clinical applications? Yes, it does.


Two Meanings of “Vigorous” Exercise and Their Effects on Whole-Bone Strength

(Frost,1998a, 1999b)

These meanings could have special importance in space, sports, and physical medicine, and in rehabilitation, geriatrics, biomechanics, and pharmacology.

To explain, muscle power and neuromuscular coordination help to achieve excellence in many sports, but at present it seems whole-bone strength adapts chiefly to peak momentary muscle forces. Thus, low-force muscle contractions repeated to exhaustion, as in marathon or treadmill running, or in long distance walking, swimming, and bicycling, can increase muscle endurance but not momentary muscle strength or whole-bone strength (Micklesfield et al., 1995). However, maximal-force muscle contractions, as in weight lifting or sports like soccer (Wittich et al., 1998) that involve violent accelerations of the body— “supranormal” has that meaning here— can increase momentary muscle strength and put much larger loads on bones than low-force exercises like those above. Note that muscle strength can increase faster than whole-bone strength (Heinonen et al., 1995).


Weight lifters and soccer players should have greater bone strength than devotees of low-force exercises and they do (Frost, 1990a; Karlsson et al., 1993; Marcus et al., 1996; Riggs and Melton, 1995; Smith and Gilligan, 1989; Taafe et al., 1995). Because the remodeling threshold lies well below the modeling threshold (Fig. 3), low-force exercises could still cause large enough strains to make or help to make conservation-mode remodeling keep the existing bone strength. It seems they do (Frost, 1999b, 2000a; Smith et al., 1989)**.

Physical Exercise and Whole-Bone Strength in Children, Adults, and Young Athletes

(Frost,1999bNilsson et al.,1978Sumner and Andriacchi,1996)

These matters could have special importance in pediatrics, in space, sports, and physical medicine, and in anthropology, pharmacology, geriatrics, and biomechanics. Besides aging effects on those matters (Stanulis-Praeger, 1989), biomechanical effects would occur too.

Increasing body weight and muscle strength keep increasing the size of the loads on a child's bones, so the sluggish modeling could lag behind in making bones strong enough to keep strains from exceeding the modeling threshold (Frost and Jee, 1994). That was called the adaptational lag (Frost, 1997c). When body weight and muscle strength plateau in young adults, modeling could “catch up,” reduce strains below that threshold and turn off. Declining muscle strength in most aging adults should put bones adapted to young-adult muscle strength into gradual partial disuse. That could downshift bone strains to the remodeling threshold and cause slow losses of bone next to marrow.


More vigorous exercise should more readily increase bone strength in children and young athletes than in aged subjects. That is true (Tsuji et al., 1996). Also, in aged subjects such exercise could cause large enough strains to limit further bone losses but not large enough strains to make modeling increase bone strength. That does happen (Marcus et al., 1996; Smith and Gilligan, 1989; Smith et al., 1989). That emphasizes the value of regular exercise to increase bone strength during growth, and hopefully help to maintain it in order to minimize fractures in aging adults (Schönau et al., 1998). The above “adaptational lag” should increase fractures during our adolescent growth spurt but let them decrease in young adults. Both of these things occur (Frost, 1997c; Rockwood and Green, 1991; Wiley and McIntyre, 1980). In microgravity conditions in orbit, exercising against maximal resistance might minimize bone losses more effectively than treadmill running or riding a stationary bicycle (Rittweger et al., 1999), although this idea has not been tested yet.

Fracture Patterns of the Radius

This “natural experiment” offers insight into fracture patterns in general. It could hold special interest for pediatricians, gerontologists, orthopaedic surgeons, osteoporosis experts, biomechanicians, and physiologists.

In children, radius fractures from falls can affect both the diaphysis (shaft) and the metaphyseal region, but in aged adults falls usually only fracture the metaphyseal region (the wrist) (Rockwood and Green, 1991, 1997). While nonmechanical explanations were suggested for that difference (Deng et al., 2000), the above physiology offers a plausible biomechanical explanation too (granted: “plausible” does not automatically mean “correct”).

Consider that in children the radius would adapt its strength to increasing loads from growing voluntary muscle forces. Its diaphysis would adapt to combined uniaxial compression, bending, and torsional loads from muscles**, but the very low friction of the radiocarpal joint would make the metaphyseal region of the radius carry and adapt mainly to uniaxial compression muscle loads**. In young adults both parts of the radius would have adapted to such loads.

Falls on the outstretched hand can put momentarily large combined bending, torsional, and compression loads on the whole radius. In aged adults, its diaphysis would have adapted to such loads, but its metaphysis would have adapted mainly to compression loads. As a result, the bending forces from such falls would more likely fracture the metaphysis than the diaphysis. Hence the common Colle's fracture in aging adults.


Similar things could help to explain why falls in aging adults seldom fracture the femoral, humeral or tibial diaphyses**. Instead, they usually fracture the metaphyseal regions of those bones (which include the hip [femoral neck, greater and lesser trochanters, and intertrochanteric region], surgical neck of the humerus, and malleolar regions of the ankle) (Ferretti et al., 1995).

Whole-Bone Strength in Obesity

This matter would have special importance for internists, endocrinologists, metabolic bone disease, nutrition authorities, and anthropologists (Nishizawa et al., 1991). To explain, body weight provides a resistance muscles must overcome to let us work and play on earth (Ferretti et al., 1998a; Martin et al., 1998). Ergo, to pursue similar physical activities obese people would need stronger muscles than less heavy slender people. The stronger muscles would put larger loads on bones, to which the above physiology should respond by increasing bone strength**, even if nonmechanical factors help to do it. That seems to be the case (Riggs and Melton, 1995; Nishizawa et al., 1991). Presumably for such reasons most bed-ridden or otherwise chronically very inactive obese people lose bone, even when their obesity increases (Frost, unpublished observations).

Table 4 lists conversion factors for English and metric units of measure, and some stress-strain conversions for healthy lamellar bone.

Table 4. Conversion factors and symbols for units
  • a

    Values to two or three place accuracy. Taken from Frost (1998a,)Yamada (1970), and The Merck Index, 11th ed. (1989)

Symbols for units
N = Newtonmpa = megapascalpsi = pounds per square inch
kg = kilogramM = metermm = millimeter
cm = centimeterin = inchlb = pound
Approximate strain-stress equivalents for normal lamellar cortical bonea(loaded in compression parallel to the grain)
50–100 microstrain corresponds to ≈ 1–2 mpa, ≈ 140–280 psi, 0.2 kg/mm2
1,000 microstrain corresponds to ≈ 20 mpa, ≈ 2,800 psi, ≈ 2 kg/mm2
3,000 microstrain corresponds to ≈ 60 mpa, ≈ 8,500 psi, ≈ 6.1 kg/mm2
25,000 microstrain corresponds to ≈ 120 mpa, ≈ 17,000 psi, ≈ 12.2 kg/mm2
Bone's ultimate strength is a range centered near 25,000 microstrain, 120 mpa, or 17,000 psi
Some English-metric conversionsa
1 kg = 2.2 lb = 9.8 N. 1 N = 0.225 lb = 0.102 kg. 1 million N = 224,000 lb.
1 mpa = 1 million N/M2 = 145 psi = 1 N/mm2 = 0.102 kg/mm2. 1 kg/cm2 = 14.2 psi.
1 kg/mm2 = 9.8 mpa = 1,420 psi. 120 mpa = 17,400 psi = 12.2 kg/mm2. 60 mpa = 8,700 psi = 6.1 kg/mm2. 20 mpa = 2,800 psi = 2.04 kg/mm2. 1 M2 = 1,550 in2.
1 in2 = 6.45 cm2.

Some Clinical Features That Depend on Bone Modeling

(Frost,1995; Jee,1989; Schönau,1996)

These matters could have special importance for physiologists, anatomists, anthropologists, pediatricians, geneticists, internists, dentists, pathologists, histologists, pharmacologists, cell and molecular biologists, and histomorphometrists.

Some modeling functions.

Modeling formation drifts create our initial supplies of cortical bone (Jee, 1989). Modeling can slowly increase bone strength by increasing bone “mass” and reshaping a bone as in Figure 1B. Aided by the modeling threshold it sets the upper limit on a bone's strength relative to the size of the loads the bone carries. Over time periods of months or, in large bones, even years, it reshapes and strengthens an initial fracture callus or a healing bone graft to provide enough strength to endure voluntary activities (where “enough strength” means keeping strains from exceeding the modeling threshold, and satisfying Proposition 1). In such ways, modeling helps to provide the greatest strength with the least amount of material (Currey, 1984), and it affects whole bones and individual trabeculae. It might strengthen the bone supporting load-bearing implants, if the bone is alive and if its strains exceed its modeling threshold but stay below its microdamage threshold (Frost, 1992b)**. Normal modeling makes bones strong enough to minimize microdamage, fatigue failures, and the true osteoporoses described in Classifying “Osteopenias” and “Osteoporoses.”

Some modeling disorders.

These can make bones fail to satisfy Proposition 1. That failure helps to increase bone fragility in osteogenesis imperfecta** (Damjanov and Linder, 1996; Frost, 1987a; Seeforf, 1949; Sillence et al., 1979). Curiously, so far no research studied how an abnormal Type I collagen could cause the modeling and remodeling disorders that chiefly reduce bone strength, increase bone fragility and let spontaneous fractures occur in this disease (Frost, 1987a; Jaffe, 1972)**. An analogous modeling malfunction should help to cause the true osteoporoses described in Classifying “Osteopenias” and “Osteoporoses”** in which the affected bones do not satisfy Proposition 1 (Frost, 1997a; Marcus et al., 1996). Decreases or failures of modeling to make healing fractures, bone grafts, osteotomies, and arthrodeses strong enough to carry voluntary loads can cause late but uncommon failures of that healing (Frost, 1998c). A clinical clue to such a late failure: Initially the bone heals well enough to let function resume, but later the healed region develops a stress fracture or begins to angulate (Frost, unpublished observations). Excessive periosteal formation drifts in Paget's disease and congenital lues cause many of the bone deformities associated with those disorders (Jaffe, 1972; Luck 1950). Inability to form woven bone is lethal for mammals (Dickman, 1997), but not for elasmobranchs like sharks and skates, which only have cartilage in their skeletons. Most laminar periosteal reactions called “periostitis” by radiologists represent new formation drifts of woven bone, in reaction to some local pathology such as a stress fracture, an inflammatory process or a neoplasm in the bone. Sometimes humoral agents can cause them, as in pulmonary hypertrophic osteoarthropathy and scurvy (Damjanov and Linder, 1996; Jaffe, 1972; Luck 1950), and in the formation drifts incited by systemically administered prostaglandin E-2 (High, 1988; Tang et al., 1997).

Some Clinical Features That Depend on Bone Remodeling

These matters could also have special importance for physiologists, anatomists, anthropologists, pediatricians, internists, geneticists, dentists, pathologists, histologists, pharmacologists, cell and molecular biologists, and histomorphometrists.

Some remodeling functions.

Remodeling replaces primary spongiosa beneath growth plates with the secondary spongiosa made of lamellar bone (Frost and Jee, 1994; Jee, 1989). It helps to replace mineralized cartilage in osteochondromas with a normal secondary spongiosa (Jaffe, 1958). Aided by the remodeling threshold, remodeling sets the lower limit on whole-bone strength, and thereby helps to determine the width of a bone's adapted window (AW) in Figure 3. It replaces fracture callus with lamellar bone. It repairs limited amounts of microdamage (Mori and Burr, 1993). In its disuse mode, it removes mechanically unneeded bone next to marrow (Frost, 1998b). Presumably that explains the nearly total lack of spongiosa in postnatal diaphyseal marrow cavities, and the loss of spongiosa and expansion of the marrow cavity diameter in all adult-acquired osteopenias**. Disuse-mode remodeling of bone next to marrow causes a woman's normal postmenopausal bone loss**. Where woven bone carries postnatal loads, remodeling usually replaces it with lesser amounts of lamellar bone. Remodeling has a minor role in homeostasis (see Homeostasis and Bone). Acute disuse increases BMU creations and bone turnover by remodeling, while increased mechanical usage tends to depress those creations and turnover**. Still, it seems increased microdamage during suddenly increased mechanical usage can override the latter effect and independently increase BMU creations and bone turnover (Frost, 1992a; Martin, 2000; Martin et al., 1998).

Some remodeling disorders.

These can fail to replace fracture callus with lamellar bone to cause some healing problems of fractures, autografts, allografts, xenografts, osteotomies, and arthrodeses (Frost, 1998c)**. That same failure impairs bone healing in osteopetrosis (Bollerslev, 1989; De Palma et al., 1994). Failure to replace primary spongiosa with secondary spongiosa causes one kind of osteopetrosis (Jaffe, 1972). Combined with modeling malfunctions, disuse-mode remodeling would help to cause all true osteoporoses including osteogenesis imperfecta**, and it (not osteoclasts alone) seems to cause all adult-acquired osteopenias on earth and in orbit**. Disuse-mode remodeling helps to cause loss of femoral calcar bone after some total hip replacement arthroplasties (Frost, 1992b; Pritchett, 1995), and it (not osteoclasts alone) causes the bone loss associated with treatment with adrenalcortical steroid analogs like Prednisone**. It helps to cause subchondral cysts in osteoarthritis, and may help to cause some lytic bone lesions associated with things like sarcoid, multiple myeloma, some kinds of bony metastases, unicameral bone cysts, and giant cell tumors of bone (Jaffe, 1958, 1972). Antiremodeling agents like estrogen and the bisphosphonates depress disuse-mode remodeling (not just osteoclasts) and help to retard local and generalized bone losses (Fleisch, 1995; Frost, 1997a). Impaired microdamage repair by BMUs causes or helps to cause osteochondritis dissecans, aseptic necroses of bone, and spontaneous fractures of irradiated bone (Frost, 1986), as well as stress fractures in athletes and military trainees, pathologic fractures, pseudofractures in osteomalacia, and spontaneous fractures in true osteoporoses (Frost, 1989a)**. That impaired repair also helps to explain the loosening of some internal fixation implants and some load-bearing endoprostheses (Frost, 1992b).


Modeling and remodeling may have other as yet unrecognized functions and disorders. A special bone resorption mechanism that remained unstudied after its original report may also participate in some bone disorders (Jaworski et al., 1972). While woven bone can form de novo, meaning where no bone of any kind existed before, lamellar bone only forms on preexisting bone of any kind (Frost, 1986).

Classifying “Osteopenias” and “Osteoporoses”

This could have special importance in metabolic bone disease, absorptiometry, radiology, internal medicine, endocrinology, geriatrics, genetics, anthropology, nutrition, pathology, space medicine, pharmacology, and histomorphometry.

In the 1990s, participants in WSS Jee's seminal Hard Tissue Workshops (Jee, unpublished data2) suggested the physiology summarized in Summary of the New Physiology above could cause four kinds of “osteoporosis” that could have similar bone “mass” deficits, and thus similar Z scores (Frost, 1997a). They do occur and some of their clinical features were known for over 40 years (Riggs and Melton, 1995; Snapper, 1957; Urist, 1960). Those participants suggested the following names.

In physiologic osteopenias, chronic muscle weakness and physical inactivity would make normal modeling and remodeling cause a corresponding disuse-pattern osteopenia in which voluntary activities and loads on bones do not cause spontaneous fractures. Here bones would satisfy Proposition 1, and only injuries like falls cause fractures, usually of extremity bones like the hip and wrist (Lauritzen, 1996). As Runge et al. (2000) and Overstall et al. (1997) noted, impairments of muscle strength, coordination, balance, and vision help to increase falls and fractures in aging adults. These osteopenias can affect children, women, men, most aged adults, and most persons with chronic muscle weakness and/or debilitating illnesses (Table 3)**. Presumably the loss of bone in women going through menopause also causes such an osteopenia (Christiansen et al., 1981), since over two-thirds of such women never develop spontaneous fractures. In former times and in older people, these osteopenias were often called “senile osteoporoses.”

Table 3. Some conditions that cause chronic disuse and muscle weakness in humans (and related osteopenias)*
  • *

    In causing an osteopenia, the relative importance of the mechanical disuse and muscle weakness, and of the biochemical-endocrinologic abnormalities accompanying some of these entries, is still uncertain. So far, few studies tried to quantify the muscle and mechanical usage effects. The Utah paradigm suggests the mechanical effects would dominate most biochemical-endocrinologic ones. RA: rheumatoid arthritis.

AsthmaEmphysemaPulmonary fibrosis
Renal failureHepatic failureCardiac failure
Metastatic cancerDepressionStroke
Muscular dystrophyMultiple sclerosisAlzheimer's disease
Organic brain syndromeHuntington's choreaMyelomeningocele
Lou Gehrig diseaseParalysesLeukemia
Cystic fibrosisStill's diseaseAlcoholism
Drug addictionNursing home residenceJuvenile RA

In true osteoporoses, still-enigmatic modeling and remodeling malfunctions cause a disuse-pattern osteopenia in which voluntary activities and muscle forces do cause spontaneous fractures. Here the affected bones do not satisfy Proposition 1. Much less common than physiologic osteopenias, these osteoporoses include in part osteogenesis imperfecta, hyperphosphatasia, and idiopathic juvenile osteoporosis, in which the spontaneous fractures can affect both the spine and extremity bones (Dimar et al., 1995; Marcus et al., 1996). A more widely discussed kind affects women more than men and seldom affects children (Riggs and Melton, 1995). Its spontaneous fractures affect thoracic and lumbar vertebrae but, curiously, rarely affect the pelvis and extremity bones (the still-debated issue of how to classify spontaneous vertebral “fractures” is not discussed here; Eastell et al., 1991; Marcus et al., 1996). Presumably, it also involves excessive microdamage accumulations (Heaney, 1993; Vernon-Roberts and Pirie, 1997). Here too the osteopenia facilitates extremity-bone fractures from falls. In former times these were often called “symptomatic osteoporoses.”

In combined states features of those two affections seem to combine in various ways**.

In transient osteopenias, a regional disuse-pattern osteopenia occurs after a fracture, burn, or other severe injury. Two or more years after the injury heals and physical activities resume, the affected bones regain the strength needed to endure voluntary activities for the rest of life**. In proof, late refractures of such fractures are rare (Frost, unpublished data). It seems the associated mechanical disuse and an accompanying regional acceleratory phenomenon cause this “naturally reversible osteopenia”** as Z.F.G. Jaworski dubbed it (Jaworski, 1984). Since spontaneous fractures do not occur in it, it should constitute a physiologic bone response to an injury (Garland et al., 1994). In former times these were sometimes called “posttraumatic osteodystrophies.”


X-ray absorptiometry cannot distinguish those four conditions from each other, nor can it alone evaluate bone health**. Defining “osteoporosis” by BMD-derived Z scores (Kanis, 1994; Marcus et al., 1996) could need revision or supplementation, since that could suggest that “osteoporoses” and “osteopenias” are only different severities of the same thing, like the hemoglobin in mild and severe pernicious anemias**. Yet as defined above, they differ biologically, pathologically, and pathogenetically. We need new standards for the muscle strength–bone strength relationship** (Ferretti et al., 1998b; Schiessl et al., 1998; Schönau et al., 1998), and will need to learn more about muscle itself (Dickinson et al., 2000; Worton, 1995). Muscle weakness plus impairments of balance (Schroll et al., 1999), neuromuscular coordination, and vision cause most of the falls that, in turn, cause most extremity bone fractures (so-called “osteoporotic fractures”) in aging and aged humans. The literature shows growing recognition of that fact (Lauritzen, 1996; Nguyen and Eisman, 2000; Runge et al., 2000; Tinetti et al., 1994), so future “risk of fracture” studies might try to account for it. Increased exercise and/or increased muscle strength [perhaps due to exercise, androgens (Bhasin et al., 1996) or growth hormone (Ogle et al., 1994; Ullman and Oldfors, 1989)] should help physiologic osteopenias, but they could make true osteoporoses worse (Frost, unpublished data). If so, it would be imperative to distinguish between those disorders before prescribing or advising such things. Agents that could turn modeling on, could normalize and cure an osteopenia**, while agents that could turn disuse-mode remodeling off could prevent one or progression of an existing one**. Yet we need better agents to do such things than the currently available bisphosphonates, parathyroid hormone, prostaglandins, and estrogens (Harris et al., 1996; Ma et al., 1994; Takahashi et al., 1991; Wronski et al., 1989). Searches for intrinsic bone disorders, including genetic disorders in bone cells, that could cause physiologic and transient osteopenias should be futile, since the chief cause of the former would be muscle weakness (which of course could depend on genetic factors), and of the latter, trauma. In my experience, physiologic osteopenias outnumbered true osteoporoses by more than five to one. In the past, did we exaggerate the need for abundant dietary calcium to minimize or prevent osteopenias, osteoporoses and fractures (Bronner, 1994; Gallagher, 1990; Nordin and Heaney, 1990; Recker and Heaney, 1985)? The bone loss in microgravity situations should exemplify a disuse-pattern osteopenia caused by greatly reduced muscle loads on bones. At least in my view, the nonmechanical explanations proposed are erroneous. Searches for genetic errors in bone that might explain “osteoporosis” as diagnosed by Z scores (Kanis, 1994) should be futile in the cases of physiologic osteopenias (since their cause would usually lie in muscle) and transient osteopenias (which should represent normal responses to trauma). Seeking other bone effector-cell disorders that would cause physiologic osteopenias should be futile too, and so should seeking the cause of spontaneous fractures only in the spine by studying unaffected bones like the ilium and tibia. Physiologic osteopenias should only affect hollow bones with marrow cavities, which seems to be true (Frost, unpublished data).

Noninvasive Absorptiometric Evaluation of Whole-Bone Strength

This matter could have special importance in metabolic bone disease, in absorptiometry by X-ray, magnetic resonance imaging, and/or ultrasound, in radiology, and in “osteoporosis”-related research.

To explain, material in classifying “Osteopenias” and “Osteoporoses” indicates the need to evaluate whole-bone strength noninvasively in patients. Yet no current absorptiometric method can reliably evaluate a bone's materials properties or its microdamage burden. As for the other two factors in whole-bone strength (see Four Physcial Features Combine to Determine Whole-Bone Strength, above), bone mineral “density” (BMD) and content (BMC) measured by dual energy X-ray absorptiometry (DEXA) became popular ways to evaluate the “mass” factor, but they cannot distinguish between woven, plexiform, and lamellar bone (Jiang et al., 1999; Kanis, 1994). Also, “mass” factors alone do not indicate whole-bone strength reliably (Banu et al., 1999)**, and neither do current ultrasound methods** (Ferretti et al., 1998b; Nielsen, 2000; van der Perre and Lowet, 1996).

However, Bone Strength Indices (BSIs) obtained by peripheral quantitative computed tomography (pQCT) that account for both the “mass” and architectural factors can provide much better estimates of a bone's true strength (Banu et al., 1999; Ferretti, 1997, 1999; Ferretti et al., 1998a,b; Jiang et al., 1999; Schiessl and Willnecker, 1999; Wilhelm et al., 1999), so they should see increasing use in the future. Please note that the major issue in life seems to be bone health, which in my view would constitute the relationship between a bone's strength and the size of the voluntary loads it carries and would normally adapt to.


Evaluating bone health would require comparing a bone's strength to the usual loads on it, and then comparing that relationship to suitable norms. Because the largest voluntary loads come from muscles, that would require comparing bone strength to muscle strength.

Ferretti et al. (1989), Schiessl and Willnecker (1999), and Schönau et al. (1998) have begun to obtain such “bone strength-muscle strength” norms. No current method of bone absorptiometry can by itself evaluate a bone's health as Proposition 1 defines it. Nor can any such method distinguish the four conditions described in Classfying “Osteopenias” and “Osteoporoses” from each other, nor can the T and Z scores currently used in such work (Kanis, 1994). Such distinctions would require adding to bone strength data, further information obtained from X-rays, clinical facts, and muscle strength data.

Design and Use of Load-Bearing Implants

This matter would have special importance for orthopaedic surgeons, dental and maxillofacial surgeons, biomedical engineers, biomechanicians, and implant manufacturers. The following paragraphs concern only one of the problems such implants have (Bauer and Hirokawa, 1995; Doyle, 1993; Hamilton and Gorczyca, 1995; Jasty et al., 1994).

To explain, the above physiology suggests the design of load-bearing endoprostheses should 1.) keep typical peak strains in the supporting bone below its microdamage threshold, but 2.) let them exceed its remodeling threshold (Frost, 1992b)**. Strains in the mild overload window (Frost, 1992a) in Figure 3 might help modeling to strengthen the supporting bone, and should keep disuse-mode remodeling from removing it. These criteria should apply to load-bearing artificial joints, partial bone replacement endoprostheses, dental implants, and some spinal instrumentation.

While a bone microdamage threshold was suggested in 1983 (Burr et al., 1983) and verified later (Carter, 1984; Pattin et al., 1996), even in 2000 AD no marketed load-bearing skeletal implant intentionally tried to satisfy those two criteria. Yet it seems Branemark's dental implant system does it unintentionally, which should prove it can be done (Branemark, 1988).

As for other kinds of implants, including ones used for internal and external fixation, very osteopenic bones with thin cortices and reduced amounts of spongiosa would need more and/or larger screws, pins, and other devices to provide larger load-bearing bone-implant interfaces (Okuyama et al., 1995). Combined with suitable postoperative management, that could help to keep the unit loads on those interfaces below bone's microdamage threshold, which in stress terms seems to lie in the region of ≈ 60 megapascals. Otherwise, accumulating fatigue damage in the bone supporting the implants could and often does loosen them before satisfactory healing occurred.

Implications for Bone Healing in Fractures, Bone Grafts, Osteotomies, and Arthrodeses

These implications could have special importance for orthopaedic surgeons, pathologists, and pharmacologists, for cell and molecular biologists who study hard tissue healing, and for the designers of internal and external fixation devices. Of course, this healing poses other clinical and basic science problems, too.

To explain, in earlier views bone healing comprised a single indivisible process, and its supposed key players, osteoblasts, were aided by things like angiogenesis, apoptosis, chondroblasts, and stem cells (Aho et al., 1994; Burchardt, 1983; Brand and Rubin, 1987; Habal and Reddi, 1992; Hall, 1991; Luck, 1950; Rahn, 1982; Rhinelander and Wilson, 1982; Sherman and Phemister, 1947).

However, its omissions make that view suspect. The true key players in that healing include four tissue-level phases, the callus, remodeling, and modeling phases, accompanied by a regional acceleratory phenomenon (RAP)** (Frost, 1998c; Woodard, 1991). Each phase can malfunction independently of the others**, so many different kinds of healing problems could and do occur that do not stem from known treatment errors. While former anatomists, histologists, and pathologists described the light-microscopic tell-tales of those things quite well (Gegenbaur, 1867; Lewis, 1906; Putschar, 1960; Weinmann and Sicher, 1955), their functional significance in this matter remained unknown until the Utah paradigm gelled.

Initially a soft fracture callus forms with new vessels, supporting and precursor cells, osteoblasts making woven bone, and often chondroblasts making hyaline cartilage. It fills the gaps and surrounds, embeds, and welds to the fragments of the fracture or graft, and it lacks a general “grain” (Weinmann and Sicher, 1955). After the callus mineralizes remodeling BMUs begin to replace it and/or any graft material with packets of new lamellar bone, the “grain” of which usually parallels the largest local compression and tension strains. Presumably guided by those strains and partly overlapping “2,” modeling begins to modify the shape and size of the callus to make it strong enough to satisfy Proposition 1. Those three phases last longer in adults, large bones, and diaphyses than in children, small bones, and metaphyses. A fracture, arthrodesis, osteotomy, or grafting operation normally incites a RAP (Garland et al., 1994) that lasts throughout the healing process and accelerates the “1,2,3” phases**. Otherwise, a delayed union or a “biologic failure” of healing can ensue (Frost, 1989b). Besides impaired regional blood supply, sensory denervation as in some diabetics increases the likelihood of an inadequate RAP, which nevertheless seldom happens in children (Frost, 1986). The possibility that cigarette smoking might impair a RAP, and thus bone healing too, seems to deserve study (Cook et al., 1997). The osteoclast defect that causes osteopetrosis impairs replacement of fracture callus with lamellar bone (“B” above) (de Palma et al., 1994), which should help to explain impaired bone healing in that disease**.

In my experience, most impairments of bone healing not due to treatment errors stemmed from disorders in the “1” and “4” phases, the “4” disorder being the most common.

Role of Strain.

Mounting evidence indicates that small strains help to guide the remodeling and modeling phases of bone healing (Blenman et al., 1989; Carter et al., 1988; Claes et al., 1994; Frost, 1989b; Hanafusa et al., 1995; Kenwright and Goodship, 1989; Mosely and Lanyon, 1988; Wolff et al., 1981). Lacking any strains, disuse-mode remodeling tends to remove the callus, modeling stays off, and healing can retard or fail** (see “disuse” in the Glossary). All orthopaedists know that excessive strains (gross motion) can prevent healing. The “permissible” strains might lie in the 100–2,000 microstrain region. For comparison, bone's fracture strain is a range centered near ≈ 25,000 microstrain (Currey, 1984; Reilly and Burstein, 1991). The 100–2,000 microstrain span includes the adapted and mild overload windows in Figure 3 (Frost, 1992a). In compliant (i.e., not yet rigid and strong) healing fractures, bone grafts, or arthrodeses including spinal fusions, very small loads could cause harmfully large strains.

Cell and Molecular Biology on Which “1–4” Should Depend.

Bone healing should also depend on the growing numbers of known humoral and molecular-biologic influences on bone cells. The humoral influences include in part hormones, vitamins, minerals, and drugs (Bak et al., 1991). The molecular-biologic influences include in part cytokines, growth factors, other ligands, angiogenesis, apoptosis, stem cell hierarchies, “supporting cell” functions, cell proliferation and differentiation, and gene expression mechanisms and patterns (Barnes et al., 1999; Caplan and Dennis, 1996; Goldring and Goldring, 1996; Gowen, 1992; Manolagas and Jilka, 1995; Ridley, 2000; Urist, 1995). One might add electrical treatment to that list (Lavine and Grodzinski, 1987). So far, a lack of appropriate studies leaves us uncertain about how such things would affect the true key players in this healing, the tissue-level “1–4” phases.

According to the Utah paradigm, very similar observations would apply to the healing of fascia, ligaments, tendons, and articular cartilage (Frost, 1995).

Homeostasis and Bone

This matter would have special importance for physiologists, internists, endocrinologists, and specialists in nutrition.

In early views, the responses of osteoclasts and BMUs to parathyroid hormone, calcitonin, and other humoral agents were viewed as essential for calcium homeostasis (Albright and Reifenstein, 1948; Barzel, 1970; Favus, 1999; Rasmussen and Bordier, 1974; Snapper, 1957). Some people even viewed that as their chief function.

Yet bone can handle most homeostatic challenges without BMUs or osteoclasts (Frost, 1986)**. In proof, problems with homeostasis, tetany, and acid-base physiology seldom occur in osteopetrosis, where osteoclasts function poorly or not at all (Bollerslev, 1989; Key and Ries, 1996). Also, in dogs large doses of a bisphosphonate suppressed osteoclastic and osteoblastic activities for 9 months without causing hyper- or hypocalcemia, tetany, or disturbed acid-base physiology (Flora et al., 1981). Since those doses did cause spontaneous fractures that did not heal until the treatment stopped, it became necessary to ensure such agents do not do similar things in humans (Fleisch, 1995). Parenthetically, while preclinical studies of some bisphosphonates claimed they did not increase microdamage in bone, a later study could suggest otherwise (Mashiba et al., 2000).

At least three other mechanisms that do not involve osteoclasts help in the homeostatic function of bone and handle most homeostatic challenges very well (Frost, 1986; Norimatsu et al., 1979)**. One of them could involve an osteocyte-based mechanism originally proposed by Arnold et al. (1971) and later supported by studies by, among others, Borgens (1984), Tate et al. (1998), and Rubinacci et al. (2000). However, in prolonged calcium deprivation or malabsorption, disuse-mode remodeling probably can help to maintain homeostasis.

Role of Nutrition

In former views, adequate nutrition dominated the development of healthy and strong bones (Bronner, 1994; Heaney, 1990; McLean and Urist, 1961; Kuhlencordt and Bartelheimer, 1981; Tylavsky and Anderson, 1988; Vaughn et al., 1975). While serious malnutrition certainly can affect bone strength adversely, and muscle strength too (Shires et al., 1980), in the Utah paradigm most things like protein, calcium, vitamins, and calories would act mainly like the fuel and engine in a car. Without them a car cannot move, but they do not drive it. Instead, the car's driver does that. For the bone “car,” the “driver” seems to be mechanical usage, muscle strength, and the related bone strains instead of nutritional factors**. In proof, no nutritional supplements can make sedentary people develop the strong bones of weight lifters, nor can they normalize whole-bone strength in paralyzed limbs (Frost, unpublished data). Furthermore, supplemental dietary calcium seems to have little effect on bone “mass” in normal children (Lee et al., 1996).


Three Caveats

Besides dynamic longitudinal bone strains, other things could help to control modeling and remodeling. They include shear, strain gradients (Frost, 1993; Gross et al., 1979; Judex et al., 1997), and strain rates, frequencies and repetitions, and other things too (Evans, 1957; Lanyon, 1996; Martin et al., 1998; Mosely and Lanyon, 1988; O'Connor et al., 1982; Rubin and McLeod, 1995). Until those things are resolved, longitudinal strains can provide reliable indicators of the loads on bones. Ergo, where this text mentions strain as a control of a biologic activity, “or equivalent stimulus” is always understood.

Bone physiology combines anatomical and biomechanical concerns with subjects like endocrinology, biochemistry, cell biology, and homeostasis. The resulting amalgam has important mechanical functions, but poor interdisciplinary communication left people in many fields unaware of that amalgam and its applications (Brown and Haglund, 1995; Parfitt, 1997). As one result, they tried to explain bone's anatomy, physiology, and clinical disorders with generally accepted earlier views about independently working effector cells. In retrospect, they may have tried to explain too much with too little.

As another result, when those early views met the newer ones in this text, controversies began that only time and help from many people can resolve. But controversies fuel progress in all science, so why not air, instead of discourage, any about issues raised by the newer physiology?

On What Cellular, Molecular-Biologic, Genetic, and Pharmacologic Roots Does the Above Physiology Depend?

It must depend on such roots, but the post-1950 rush to study bone's effector cells pretty much overlooked that. For recent reviews about those cells see Caplan and Dennis (1996), Goldring and Goldring (1996), Duncan and Turner (1995), Mundy (1996), Parfitt and colleagues (1993, 1995, 1996), Raisz (1988), Rodan (1997), and Turner et al. (1994).

Yet growth hormone and somatomedins (Inzucchi and Robbins, 1994; Kalu et al., 2000), androgens (Bhasin et al., 1996), cortisone analogs, vitamin D, calcium, and genes all affect muscle strength and could indirectly affect bone strength in that way (Dickinson et al., 2000). These and other factors might also potentiate the mechanostat's responsiveness to mechanical and other influences to affect whole-bone strength in that way too (Frost and Schönau, 2000); recent studies support that idea (Gasser, 1999; Halioua and Anderson, 1989; Jee, 1999; Tang et al., 1997; Yao et al., 2000). Some factors might even affect the impaired balance and neuromuscular coordination that help to cause the falls that, in turn, cause most extremity bone fractures in aged adults (Guralnik et al., 1995; Runge, 1997; Steinhagen-Thiessen and Borchelt, 1996). While few if any efforts to find genetic causes for so-called “osteoporosis fractures” studied the associated problems with balance, neuromuscular coordination, muscle strength and vision, simple, effective ways to study such things in out-patient settings do exist (Runge et al., 2000).

Consequently future studies must find how such roots support the above physiology. That may depend heavily on live-animal research, because as Parfitt (1995) and Gasser (1999) also note, bone's tissue-level mechanisms do not function normally in vitro (Frost, 1986). W.S.S. Jee's laboratory at the University of Utah pioneered ways to do such in vivo work (Jee, 1995).

Role for Controlled Vibration?

High loading rates of frequent loads with small amplitudes (including but not limited to ultrasound) may have useful effects in treating “osteoporoses,” some hard and soft tissue healing problems, and other matters (Abendroth et al., 1998; Flieger et al., 1998; Heckman et al., 1994; Wood, 1987). Among others, H. Schiessl in Germany and H. Sievanen (Sievanen et al., 1996) in Finland study the effects of such vibration on human bones, joints, muscles, and neuromuscular physiology (personal communications, 1998).

On the Cartilage-Bone Relationship

Question: Are things like congenital hip dysplasia (Stanisaljevic, 1964), genu varum, scoliosis, club foot, metatarsus varus, achondroplastic dwarfism, Madelung's deformity, Marfan's syndrome (Joseph et al., 1992), and hallux rigidus examples of bone disorders? They are not. Instead they stem from modeling disorders of the cartilage in joints and growth plates (Frost, 1995, 1999a; Frost and Jee, 1994). Usually cartilage conducts and bone plays first violin in the skeletal “orchestra” (Frost, unpublished observations), and far more often than not the statistical abnormalities in bone architecture in such conditions represent effective adaptations to loading changes caused by the cartilage problems (Frost, 1995). In proof, spontaneous fractures of such bones are rare (Frost, unpublished data).

Past, Present, Future

As a young man, my “bibles” for bone physiology included books by Jaffe (1958, 1972), McLean and Urist (1961), and Weinmann and Sicher (1955), and a chapter by Putschar (1960). Comparing their content to the above material suggests how much progress occurred in understanding bone, bones, and Wolff's Law. History suggests more progress will come (Maddox, 1999; Mayr, 1961, 2000), and many devils will show up in the details too. But the above physiology's parent, the Utah paradigm, keeps evolving to account for new evidence and ideas (and “devils”), so it could provide a kind of gold standard in such matters for years to come.


The author thanks the staffs of Pueblo's Southern Colorado Clinic, Parkview Episcopal Hospital, and St. Mary Corwin Hospital for their time producing this and related articles, and David Gavin and Ralph Scott for the drawings. Thanks also to the orthopaedic surgeons trained at Henry Ford Hospital between 1957–1973 for their spontaneous assistance in a time of great troubles. Other colleagues who have offered perceptive comments and advice regarding matters discussed in this article include: J.S. Arnold, D.B. Burr, Z.F.G. Jaworski, W.S.S. Jee, R.B. Martin, A.M. Parfitt, E.L. Radin, R.R. Recker, H.E. Takahashi, M.R. Urist, and C. Woodard.

  • 1

    Frost unpublished data: Personal observations during 50 years of experience in orthopaedic surgery, education, research and pathology, strongly supported by unpublished findings of others.

  • 2

    Jee WSS: Hard Tissue Workshops organized annually since 1965 by Professor Jee provide a uniquely seminal and multidisciplinary forum for presenting and discussing new methods, evidence and ideas about human skeletal disease. These workshops are attended by hundreds of international authorities and fellows in many disciplines, sponsored by the University of Utah and supported by private and federal funds. They have had a more profound effect on how people think about and study skeletal disease than any other meetings in this century. The Utah paradigm had its genesis at these Workshops; hence its name.



As some terms have vague or even different meanings in the medical literature, the intended meaning of some terms in this article are listed below. (*: The scientifically correct meaning in 2000 AD.)

architecture: the size, shape, and orientation of a bone, the amounts of bone tissue in it, and the arrangement of that tissue in anatomical space.

BMU: the Basic Multicellular Unit of bone remodeling (Jee, 1989)*. In 3 or more months and in a stereotyped, biologically-coupled Activation → Resorption → Formation or “ARF” sequence, it turns over ≈ 0.05 mm3 of bone. The resulting new packet of bone was called a bone structural unit (BSU) by Jaworski (1984). When a BMU makes less bone than it resorbs, that “disuse mode” tends to remove bone permanently, but only for bone next to or close to marrow. When it resorbs and makes equal amounts of bone, that “conservation mode” turns bone over without net gains or losses. Completed BMUs do not seem to make more bone than they resorb so they do not increase bone “mass.” Healthy adult humans may create and complete about 3 million BMUs annually, but in disease and other circumstances that can change more than five times (Frost, 1995).

bone “density”: since the true physical density of bone as a material varies little with age, sex, and species, “density” as absorptiometrists use the term only provides an estimate of the amount of bone in the path of one or more X-ray beams as a bone-mineral equivalent (here one can assume gamma rays and X-rays are the same). While many still think otherwise, true bone density is normal in most osteoporoses and osteopenias (Seeman, 1997). When in quotes in this article, “density” has its meaning in absorptiometry.

bone “mass”: the amount of bone tissue in a bone or skeleton, preferably viewed as a volume minus the volume of the soft tissues in the marrow cavity. In absorptiometry, it does not mean mass as used in physics. When in quotes in this article, it has the absorptiometric meaning.

DEXA: dual energy X-ray absorptiometry. Often also written as “DXA.”

disuse: bones need some criterion to recognize this. When a bone's peak strains down-shift into or below the remodeling threshold region in Figure 3, for that bone that would signal the existence of disuse, no matter how small or big the bone (Frost, 2000a). In such situations, disuse-mode remodeling usually removes bone next to marrow. “Disuse” would be the relationship between a bone's strength and the size of its usual peak loads and the strains they cause. The relationship between those strains and the remodeling threshold would provide a natural criterion for “recognizing” disuse.

disuse-pattern osteopenia: as a steady state, an osteopenia in which endocortical bone loss expanded the marrow cavity, loss of trabecular bone reduced its amount, cortical porosity remains essentially normal, and outside bone diameter does not decrease, or may even increase a bit. The reduced outside bone diameter in some children's osteopenias usually reflects failure of modeling to increase it instead of an effect of periosteal bone loss. In steady-state osteopenias, surface-referent bone tissue dynamics tend to be normal for the subject's age (Recker, 1983).

drifts: see “modeling” below, and Figure 1.

effector cells: here, differentiated osteoblasts and osteoclasts but not their precursor or other cells. The effector cells directly make or resorb bone, so by that definition osteocytes would not be effector cells.

mechanical competence: the state in which bones endure voluntary physical activities for life without developing spontaneous fractures. Sometimes called “biomechanical competence.” The antonym, “mechanical incompetence,” means the state in which voluntary physical activities (not injuries) do cause spontaneous fractures; modeling and/or remodeling disorders would usually cause it.

microdamage: microscopic physical damage in a structural material due to materials fatigue (Martin et al., 1989)*. To increase the fatigue life of inanimate structures, engineers usually add more structural material. But skeletons can detect and repair limited amounts of fatigue damage to keep it from accumulating, so they only need enough strength to keep strains below the level that could cause larger amounts. Presumably they could carry loads that cause smaller amounts indefinitely.

microdamage threshold: the strain range above which new microdamage begins to escape repair and accumulate (MESp in Fig. 3). It seems to center near 3,000 microstrain, which corresponds to a stress of about 60 megapascals. Pattin et al. (1996) found that as the loads that originally cause strains in the 2,000 microstrain range only double to cause 4,000 microstrain, the resulting fatigue damage increases over 500 times.

modeling: the biologic processes that produce functionally purposeful sizes, shapes and organization to all skeletal organs (Jee, 1989)*. Mostly independent resorption and formation modeling drifts do it in bones. Normally it fits bones to their voluntary mechanical usage to keep that usage from breaking them. That is done by making a bone strong enough to keep its typical peak strains from exceeding bone's modeling threshold.

modeling threshold: the genetically-determined Minimum Effective Strain range (or equivalent Stimulus; MESm in Fig. 3) for mechanically controlled bone modeling. Where strains exceed it modeling turns on; where strains stay below it modeling turns off. It seems to center near 1,000 microstrain in most young adults, which corresponds to a stress of about 20 megapascals (Table 4).

muscle strength: the maximum momentary contractile force exerted by a muscle can be expressed in Newtons or kiloponds (the attraction of earth's gravity for a mass of one kilogram) (Dickinson et al., 2000; Murray et al., 1980)*. Or muscle strength can be measured as the peak torque in Newton-meters produced by muscle forces across joints like the hip, elbow, knee, and fingers. That differs from endurance, which concerns how long and often submaximal muscle forces can be exerted, as in marathon running*. It differs from mechanical work or energy, which can be expressed in Newton-meters, Joules, or kilowatt-hours*. It differs from muscle power, which concerns how rapidly mechanical work is done and is usually expressed in Newton-meters/sec, Joules/sec, or watts (one Joule/sec = one watt)*. Since bones seem to adapt their strength and stiffness to the typical peak momentary loads they carry, accounting for these distinctions can minimize errors in interpreting and discussing mechanical usage effects on bone strength and “mass.”

osteopenia: here, less whole-bone strength than usual for most healthy people of the same age, height, weight, sex, and race. Also less bone strength than before in the same person. It need not represent a disease nor stem from an intrinsic bone disorder. Affected bones would break more easily. In clinical work, probably the commonest cause of an osteopenia is chronic muscle weakness. But it is currently and usually expressed in terms of reduced bone “mass” as evaluated by DEXA.

osteoporosis: defining this was debated for decades (Nordin, 1987; Urist, 1960). The currently accepted 1994 WHO “standard” for diagnosing an “osteoporosis” consisted of a bone mineral “density” or content over 2.5 standard deviations below the applicable norm (Kanis, 1994). Some suggested an “osteopenia” consists of a reduction in bone “mass” between 1.0 and 2.4 standard deviations below the applicable norm. That idea does not depend on the osteopenia's pathogenesis, yet effective treatment should depend on it. Reviews published after 1985 show many authors find the “Type I, Type II” terms confusing (Riggs et al., 1998). The pathogenetically-based terms in Part III of this text would supplement older ones. In this text, without quotes the term signifies any osteopenia in which voluntary activities (not trauma) cause spontaneous fractures. When in quotes, it would have the above absorptiometric meaning,

remodeling: turnover of bone in small packets by BMUs (Jee, 1989)*. Pre-1964 literature did not distinguish it from modeling and lumped them together as “remodeling.” While drifts and BMUs seem to create and use the same kinds of osteoblasts and osteoclasts to do their work, in different parts of the same bone at the same time the ‘blasts and ’clasts in drifts and BMUs can even respond oppositely to the same stimulus (Chen et al., 1995; Yeh et al., 1995). Since locally increased remodeling increases local bone formation, scintigrams (“bone scans”) usually show increased local uptake of the bone-seeking radioactive tracer, usually technetium.

remodeling space: increased BMU creations also increase the number of temporary holes in a bone and excavations on its surfaces. That causes a temporary bone loss called the remodeling space. It is temporary because when BMU creations return to normal, the existing holes refill with bone (because of the ARF sequence in BMUs) (Parfitt, 1980). Since increased bone formation accompanies increased remodeling and an increased remodeling space, that can help bone scans (scintograms) with radioactive bone-seeking agents to locate skeletal pathology when ordinary X-rays appear normal (Jergensen et al., 1990).

remodeling threshold: the genetically-determined Minimum Effective Strain range (or equivalent Stimulus; MESr in Fig. 3) that helps to control the switching of BMU-based remodeling between its conservation and disuse modes. When strains exceed it, completed BMUs begin to make and resorb equal amounts of bone to provide conservation-mode remodeling. When strains stay below it, completed BMUs next to marrow make less bone than they resorb to provide the disuse-mode remodeling that mainly affects trabecular and endocortical bone. This threshold may center near 50–100 microstrain, which would correspond to a tension or compression stress of ≈ 1–2 megapascals. One might also define the threshold as the region just to the left of the “adapted window” in Figure 3, which could put the corresponding threshold strain range in the 400–600 microstrain region.

resorption: different meanings of this term in the literature cause some confusion. Some authors use it to mean net bone loss, and in that sense discuss “antiresorption agents.” While often called antiresorption agents, estrogen and bisphosphonates really depress BMU creations (Fleisch, 1995; Jee, 1995). At first, that decreases global resorption, but due to the ARF sequence in the BMU, an equal decrease in global bone formation usually and eventually follows, so these are really “antiremodeling agents.” This text uses resorption to mean bone resorption by osteoclasts. It refers to net losses of bone as such and separately.

strain: the deformation or change in dimensions and/or shape caused by a load on any structure or structural material*. It includes stretching, shortening, twisting, and/or bending. Loads always cause strains, even if very small ones, and three kinds occur: compression, tension, and shear. Biomechanicians can express strain in microstrain units (millionths of a 100% strain), where 1,000 microstrain in compression would shorten a bone by 0.1% of its original length, 10,000 microstrain would shorten it by 1% of that length, and 100,000 microstrain would shorten it by 10% of that length (and break it). Strain emerges as an important signalling mechanism in controlling a skeleton's structural adaptations to its mechanical usage.

strength: the load or strain that, when applied once, usually fractures a bone (also the “ultimate strength”; Fx in Fig. 3)*. Normal lamellar bone's fracture strength expressed as a strain is a range centered near 25,000 microstrain (somewhat lower in adults and higher in rapidly growing mammals). That corresponds to a change from 100% of its original length to 97.5% of that length under compression, or to 102.5% of it under tension (Table 4).

strength-safety factor: when defined as how much stronger a bone is than needed to carry the typical largest voluntary loads on it, this factor would equal the ultimate strength divided by the modeling threshold when both are expressed as stresses. From Table 4 and in young adults, that would equal 120 mpa ÷ 20 mpa = 6. Since the modeling threshold determines the largest allowed bone strain or stress, when that threshold lies below the ultimate strength it creates the safety factor. Bones cannot foresee and adapt their strength to future injuries, so they must adapt to past and present voluntary physical activities instead, whether the activities are subnormal, normal, or supranormal.

“supranormal” exercise: here, high-force muscular activities that cause the largest momentary voluntary loads on bones. Examples include weight lifting and high-acceleration sports like soccer and US-style football.

typical peak strains: visualize a histogram that plots the sizes of a bone's strains (or loads) on the vertical axis, and on the horizontal axis the number of times dynamic strains of a given size occurred during, say, a week. The strains large enough to turn modeling on would be the largest ones in that histogram, but would comprise fewer than 0.01% of all strain events in that week. For example, each systolic pulse in the marrow cavity is a loading event on a hollow bone like the femur. In a week it would carry over 725,000 corresponding strains, of which only ≈ 50 from peak muscle forces might be large enough to reach or exceed bone's modeling threshold. The bone would adapt its strength and “mass” to those ≈ 50 events and pay little attention to all others. While some controversy affected this feature, the different bone strengths of long distance runners and weight lifters strongly suggest it is true. It would also be an obligatory effect of a modeling threshold's existence, which led me to infer that threshold's existence before in vivo strain studies verified its existence (Frost, 1964).