Perspective

On the Estrogen–Bone Relationship and PostmenopausalBone Loss: A New Model

HAROLD M. FROST

ABSTRACT

In this model of estrogen effects on bone, a postulated mediator mechanism in marrow would affect modeling andremodeling only of bone next to or close to it. That mediator mechanism could sense estrogen. In response to thathormone, it would let remodeling of bone next to marrow proceed in its conservation mode. This would minimizelosses of that bone and tend to prevent an osteopenia. But acute estrogen deficiency would make that mechanismswitch remodeling of bone next to marrow to its disuse mode. Meanwhile, conservation-mode remodeling wouldcontinue for haversian and subperiosteal bone. The resulting losses of bone next to marrow would expand marrowcavities, thin cortices, and reduce trabecular bone “mass,” but would not reduce outside bone diameters. Thatscheme could explain the osteopenia that follows natural or experimental estrogen deficiency in mammalianfemales. If so, as estrogen secretion rises in girls at puberty they should begin accumulating more bone next tomarrow. They do. Also if so, at menopause women should begin to lose that bone. They do. Those effects wouldexist in addition to known effects of estrogen on existing osteoclasts and osteoblasts. (J Bone Miner Res 1999;14:1473–1477)

INTRODUCTION

THIS ARTICLE SUGGESTS a new model for estrogen effectson bone strength and “mass” in women (when in

quotes below, “mass” has its meaning in absorptiometry).By 1950 it was known that osteoblasts make bone and os-teoclasts resorb it, and decreased estrogen secretion duringand after menopause leads to significant bone losses.(1)

That and other evidence led to the ideas that osteoblastsand osteoclasts functioned and were controlled indepen-dently, that estrogen unilaterally inhibits osteoclastic activ-ity, and that reduced estrogen at menopause removed thatinhibition and let osteoclastic activity increase and causepostmenopausal bone loss. Research did find many bio-chemical, cell- and molecular-biologic effects of estrogen onosteoclasts (and osteoblasts).(2)

Yet those ideas could not easily explain some anatomical,biological, biomechanical, and clinical evidence summa-rized next.

SOME EVIDENCE FOUND AFTER 1960

The tissue-level modeling andremodeling mechanisms

Osteoblasts alone do not control postnatal additions inbone strength and “mass,” and osteoclasts alone do notcontrol postnatal losses in bone strength and “mass.” In-stead the modeling and remodeling mechanisms do that,and each needs both osteoblasts and osteoclasts to do itswork.(3–5)

Global modeling by formation and resorption drifts canincrease but not decrease bone strength and “mass,” whileremodeling by BMUs (basic multicellular units) can turnbone over in two modes. In its “conservation mode” it turnsbone over without causing appreciable gains or losses ofbone. But in its “disuse mode” completed BMUs make lessbone than they resorb only for bone next to or close tomarrow (endocortical and trabecular bone).(6) The result-

Department of Orthopaedic Surgery, Southern Colorado Clinic, Pueblo, Colorado, U.S.A.

JOURNAL OF BONE AND MINERAL RESEARCHVolume 14, Number 9, 1999Blackwell Science, Inc.© 1999 American Society for Bone and Mineral Research

1473

ing bone losses expand the marrow cavity, reduce theamounts of spongiosa in that cavity, and reduce corticalthickness. Yet meanwhile, the outside bone diameter doesnot decrease and cortical porosity changes little.(7)

The holes and excavations created by working BMUstemporarily remove some bone. That temporary bone lossdefines the remodeling space,(3,5) which increased BMUcreations enlarge and decreased creations reduce. Normallyit equals about 4% of our bone “mass” but can exceed 20%of it. Remodeling space effects excepted, remodeling sel-dom if ever increases bone strength and “mass.”(3)

While modeling and remodeling each creates and useswhat seem to be the same kinds of osteoblasts and osteo-clasts to do its work, the activities of those cells can de-crease in modeling but increase in remodeling, or viceversa, in the same bone at the same time.(4,8,9) Thus, theseare biologically different mechanisms.

A role of BMU creations in modeling andremodeling activities and postmenopausal bone loss

Individual modeling drifts and remodeling BMUs onlyfunction for 3 or so months in humans, and they do not existat all times on all parts of all bone “envelopes” (i.e, peri-osteal, haversian, endocortical, and trabecular surfaces). In-stead, new ones are created if, when, and where they areneeded and as many and as long as needed.

The bone losses that follow acute estrogen deficiency inwomen (and acute disuse in both genders) usually take over5 years to slow down and plateau,(10) so they would dependon over 20 generations of new BMUs, each with its evenshorter-lived osteoblasts and osteoclasts. Those generationsdid not exist when the deficiency or disuse began, so con-trolling their long-term effects on bone strength and “mass”should require continually creating new ones. The kindsand numbers of cells involved in those processes, includingprecursor and supporting cells and the new blood vesselsthat nourish them, are currently uncertain.

Mechanical effects

Mechanical forces on bones deform or strain them. Di-rectly or indirectly those strains help to control modelingand remodeling effects on bone strength and “mass.”(3,5)

Where dynamic strains exceed a modeling threshold range,modeling increases bone strength and “mass” by combina-tions of changes in bone architecture and increases in bone“mass.” Where strains stay below this threshold, mechani-cally controlled modeling stays off. When strains stay belowa lower remodeling threshold range, modeling still stays offbut, remodeling space effects excepted, now disuse-moderemodeling permanently removes bone next to or close tomarrow. When strains exceed this threshold range, remod-eling tends to switch to its conservation mode to retain bonenext to marrow. Raising those thresholds should have thesame effects on bone strength and “mass” as acute me-chanical disuse. It should make disuse-mode remodelingbegin to remove bone next to marrow and turn modelingoff. When continued mechanical usage of steadily reducing

amounts of bone let bone strains rise to the higher remod-eling threshold range, conservation mode remodelingwould resume and make further bone losses tend to plateauat the new, lower level of bone strength and “mass.”

Figure 1 suggests some combined effects of modeling andremodeling on bone strength and “mass,” and the hierarchyof strain thresholds that apparently help to control them.

Bone loss after menopause and acute disuse: Wheredoes it come from?

First, during and after menopause in women, and in acutedisuse in males and females, permanent bone losses comemainly from bone next to marrow (endocortical and tra-becular bone), not from intracortical or subperiostealbone.(3,10) Indeed, while bone next to marrow is being re-moved the outside diameters of affected bones can increaseslightly,(2,4) so in the same bone at the same time periostealbone gains can accompany increased losses of endocorticaland trabecular bone. Upon adding estrogen after acute es-trogen loss, the losses of bone next to marrow decline orstop.(4,10)

Effects of estrogen deficiency and excess on bonetissue dynamics

Transient remodeling(2,4,10–14) space effects excepted, firstin acute estrogen deficiency BMU creations increase on allenvelopes. Yet disuse-mode remodeling only affects bonenext to or close to marrow; conservation-mode remodelingcontinues on the other envelopes. Second, upon giving es-trogen to such females, BMU creations, and thus bone turn-over and the remodeling space, decrease on all envelopes,remodeling returns to its conservation mode in bone next tomarrow, and it continues in that mode on the other boneenvelopes.

Such things suggest that somewhat different mechanismsmay control BMU creations on the one hand, and on theother the switching between conservation- and disuse-moderemodeling.

The plateau phenomenon

When acute estrogen deficiency makes a woman’s bonestrength and “mass” begin to decrease, they do not de-crease without limit(2) but always tend to plateau at somenew lower level, even though the hormonal deficiency per-sists.(10)

The marrow mediator mechanism and estrogen

Acute estrogen deficiency and acute mechanical disuseeach turns disuse-mode remodeling on for bone next to orclose to marrow, but not for haversian or subperiostealbone. That suggests some mechanism in marrow can senseboth that hormone and mechanical disuse. In response, thismechanism could modify modeling and remodeling of bonenext to or close to it, and only of that bone.(6) This could

FROST1474

explain why acute loss of estrogen turns disuse-mode re-modeling on only for bone next to or close to marrow. If so,increased estrogen secretion in girls at puberty should con-serve that bone better than before to let their bone “mass”increase relative to their muscle mass. That does occur (Fig.2).(15,16) Also if so, loss of estrogen at menopause shouldmake this mediator mechanism cause removal of the samebone. That occurs too.(10)

A NEW HYPOTHESIS

To account for the above evidence this model suggeststhe following features. First, the model accepts known ef-fects of estrogen and other nonmechanical agents on osteo-blasts and osteoclasts, as summarized by Riggs et al.(2) Sec-ond, but the hormone would affect the mediatormechanism in marrow too. Third, the hormone would tendto decrease BMU creations, bone turnover, and the remod-eling space on all bone envelopes. Fourth, separately, itwould make the marrow mediator mechanism turn conser-vation-mode remodeling on for bone next to marrow. Theresulting decrease in permanent bone losses there would

tend to prevent an osteopenia. Fifth, acute loss of estrogenwould increase BMU creations on all envelopes, and thusbone turnover and the remodeling space there too. Sixth,but separately, acute estrogen loss would make the marrowmediator mechanism turn disuse-mode remodeling on forbone next to marrow. The resulting increased losses of thatbone would cause an osteopenia. Seventh, estrogen’s ef-fects on BMU creations, and on the switching between con-servation- and disuse-mode remodeling, should act bysomewhat different mechanisms. Eighth, contrary to myoriginal idea,(17) estrogen would have little or no directeffect on periosteal modeling or its strain threshold range.

COMMENTS

Some predictions of the two models

The older ideas and the present model predict differenteffects of estrogen deficiency in women.

First, if estrogen deficiency increased osteoclastic activitybut had no other large effect on other bone cells: a perma-nent deficiency should eventually cause loss of all bone (forexample, in aged postmenopausal females) without an ear-

FIG. 1. Combined modeling and remodeling effects on bone strength and “mass.” The lower horizontal line suggeststypical 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; this text does not discuss microdamage). Thehorizontal axis represents no net gains or losses of bone strength or “mass.” The lower dotted line curve suggests howdisuse-mode remodeling removes bone and reduces bone strength and “mass” where strains stay below the MESr range,but otherwise keeps that bone. The upper dashed line curve suggests how modeling would increase bone strength wherestrains reach or exceed the MESm range. The dashed outlines suggest the combined modeling and remodeling effects onbone strength and “mass” (D.H. Carter originally suggested such a curve(18)). Fx 4 the fracture strain near 25,000microstrain. At the top, DW 4 disuse window; AW 4 adapted window as in normally adapted adults; MOW 4 mildoverload window as in healthy growing mammals; POW 4 pathologic overload window. (Reproduced by permission:Frost HM 1997 Strain and other mechanical influences on bone strength and maintenance. Curr Opin Orthopaed 8:60–70).

PERSPECTIVE 1475

lier plateau in bone “mass”; bone losses should affect allbones including those without a marrow cavity; bone lossesshould occur on all envelopes of those bones, since func-tionally competent osteoclasts can arise on all envelopesthroughout life.

Second, but if a signal-strength threshold in a marrowmediator mechanism controls the switching between con-servation- and disuse-mode remodeling of bone next tomarrow, and if estrogen lowers that threshold and turnsconservation-mode remodeling on for that bone, then acuteestrogen deficiency should raise that threshold. At leastfour effects should follow: at first bone losses should in-crease; transient remodeling space effects excepted, perma-nent bone losses should come from bone next to or close tomarrow; when progressively diminishing amounts of bonelet continuing mechanical usage raise bone strains to thenew threshold, conservation-mode remodeling of bone nextto marrow would resume and make bone strength and“mass” tend to plateau even though estrogen deficiencycontinues; and those bone losses should not occur in boneswithout a marrow cavity, since they would lack this media-tor mechanism. Examples include the nasal bones, turbi-

nates, ethmoids, vomer, inner ear ossicles, and the wing ofthe sphenoid.

While the new model’s predictions seem to be better,systematic studies of bone “mass” and tissue dynamics inbones without a marrow cavity are not yet available. Still,no clinical problems with such bones are known to associatewith any currently recognized kind of osteoporosis.

Some implications

The newer model suggests the needs to: find and studythe “back-up” cells, mechanisms and signals that help tocontrol the creations, work intensities and locations of mod-eling drifts and remodeling BMUs; find and study the mar-row mediator mechanism; and learn how to modify formedical needs the modeling and remodeling thresholds andthe difference between the amounts of bone resorbed andmade by the “typical” BMU.(6) The similar effects of acutedisuse and acute estrogen deficiency on bone strength,“mass,” bone-tissue dynamics, modeling, remodeling, andbone architecture suggest both challenges may act via acommon pathway.

FIG. 2. A bone-muscle mass comparison. From an Argentine study of 345 boys and 443 girls between two and 20 yearsof age,(19) this graph plots the grams of total body bone mineral content (TBMC, an index of bone strength) on the verticalaxis that correspond to the grams of lean body mass (LBM, an index of muscle strength) on the horizontal axis, asdetermined by a Norland dual-energy X-ray absorptiometry machine using dynamic filtration. Crosses, girls; open circles,boys. From left to right, each data point stands for an age 1 year older than the data point to its left, and it shows the meanbone and muscle indices for all subjects in that 1-year age group. For similar LBMs, around 12 years of age bone “mass”increased faster than before in girls, and also than in boys with similar LBMs. By ∼15 years of age, TBMC and LBM bothplateaued in girls, as shown by the closely grouped data points for their 15–20 years age groups on the far right side of theircurve. The 18–20 years age groups combine as a single data point for girls, and another for boys. Since both indices keptincreasing in 20-year-old males, they ended up with more muscle and bone than the 20-year-old girls. The data points for14- and 15-year-old girls overlap. (Reproduced by permission(15)).

FROST1476

CONCLUSION

I do not suggest that this article resolves the puzzle of themechanisms of estrogen effects on bone; instead I suspectwe only found a few more parts of a complex mechanismand problem. The matter does seem to need resolutionsince it could influence future research and the manage-ment of postmenopausal bone loss and related problems.Help from others in resolving it is invited and welcome.

ACKNOWLEDGMENTS

The author is grateful to colleagues who provided helpfulcomments and advice in the past on matters discussed inthis article. They include Profs. D.B. Burr, J.L. Ferretti,W.B. High, W.S.S. Jee, K. Kuettner, L. Garetto, A.M.Parfitt, E.L. Radin, and S. Stanisaljevic. The author is alsodeeply indebted to orthopaedic surgeons trained at HenryFord Hospital between 1957 and 1973 inclusive for theirspontaneous and generous aid in a time of great troubles.

REFERENCES

1. McLean FC, Urist MR 1961 Bone, 2nd Ed. University of Chi-cago Press, Chicago, IL, U.S.A.

2. Riggs B, Khosla S, Melton J III 1998 A unitary model forinvolutional osteoporosis: Estrogen deficiency causes bothType I and Type II osteoporosis in postmenopausal womenand contributes to bone loss in aging men. J Bone Miner Res13:763–773.

3. Frost HM 1998 Osteoporoses: New concepts and some impli-cations for future diagnosis, treatment and research (based oninsights from the Utah paradigm). Ernst Schering ResearchFoundation AG, pp. 7–57.

4. Jee WSS 1995 Proceedings of the International Conference onAnimal Models in the Prevention and Treatment of Osteope-nia. Bone 17 (Suppl ):1–466.

5. Martin RB, Burr DB, Sharkey NA 1998 Skeletal Tissue Me-chanics. Springer-Verlag, New York, NY, U.S.A.

6. Frost HM 1998 On rho, a marrow mediator and estrogen: Theirroles in bone strength and ‘mass’ in human females, osteope-nias and osteoporoses (insights from a new paradigm). J BoneMiner Metab 16:113–123.

7. Frost HM 1969 Tetracycline-based histological analysis of boneremodeling. Tissue Res 3:211–237.

8. Chen M-M, Yeh JK, Aloia JF 1995 Skeletal Tetracycline-basedhistological analysis of bone remodeling. Calcifalterations inhypophysectomized rats: II. A histomorphometric study ontibial cortical bone. Anat Rec 241:513–518.

9. Yeh JK, Chen MM, Aloia JF 1995 Skeletal alterations in hy-pophysectomizd rats: I. A histomorphometric study in tibialcancellous bone. Anat Rec 241:505–512.

10. Riggs BL, Melton LJ III 1995 Osteoporosis: Etiology, Diagno-sis, and Treatment, 2nd Ed. Lippincott-Raven Publishers, Hag-erstown, MD, U.S.A.

11. Han Z-H, Palnitkar S, Rao DS, Nelson D, Parfitt AM 1997Effects of ethnicity and age or menopause on the remodelingand turnover of iliac bone: Implications for mechanisms ofbone loss. J Bone Miner Res 12:498–508.

12. Jerome CP, Kimmel DB, McAllister JA, Weaver DS 1986 Ef-fects of ovariectomy on iliac trabecular bone in baboons. CalcifTissue Int 39:206–212.

13. Jerome CP 1989 Estimation of the bone mineral density varia-tion assoiciated with changes in turnover rate. Calcif Tissue Int44:406–410.

14. Wronski TJ, Dann LM, Scott KS, Crooke LR 1989 Endocrineand pharmacological suppressors of bone turnover protectagainst osteopenia in ovariectomized rats. Endocrinology125:810–816.

15. Schiessl H, Frost HM, Jee WSS 1998 Perspectives: Estrogenand bone-muscle strength and ‘mass’ relationships. Bone22:1–6.

16. Ferretti JL, Capozza RP, Cointry GR, Garcia SL, Plotkin H,Avlarez Figueira ML, Zanchetta JR 1998 Gender-related dif-ferences in the relationship between densitometric values ofwhole-body bone mineral content and lean body mass in hu-mans between 2 and 87 years of age. Bone 22:683–690.

17. Frost HM 1987 Bone ‘mass’ and the ‘mechanostat’: A proposal.Anat Rec 219:1–9.

18. Carter DR 1984 Mechanical loading histories and cortical boneremodeling. Calcif Tissue Int 36 (Suppl 1):S19–24.

19. Zanchetta JR, Plotkin H, Alvarez-Figueira ML 1995 Bonemass in children: Normative values for the 2–20 year-old popu-lation. Bone 16 (Suppl):393–399.

Address reprint requests to:Dr. Harold Frost

Department of Orthopaedic SurgerySouthern Colorado Clinic

2002 Lake AvenuePueblo, CO 81004 U.S.A.

Received in original form July 28, 1998; in revised form March 24,1999; accepted April 14, 1999.

PERSPECTIVE 1477