1085

Effect(s) of the DemineralizationProcess on the Osteoinductivity ofDemineralized Bone MatrixMin Zhang,* Ralph M. Powers, Jr.,* and Lloyd Wolfinbarger, Jr.*'

The relationships between residual calcium levels and particle size of grounddemineralized bone matrix and its osteoinductive potential were investigated using invitro and in vivo assays. The effects of variable residual calcium levels, variableparticle sizes, and donor age and gender were studied using a tissue culture-basedbioassay (in vitro) as well as an athymic mouse (in vivo) bioassay. The osteoinductivepotential of the bone-derived biomaterial was assessed by measuring the degree ofnew bone formation (change in percent calcium content after 4 weeks of implantation)in the in vivo assay and levels of alkaline Phosphatase activity associated with culturesof human periostea! cells (HPO cells) in the in vitro assay, respectively. Slightly de-mineralized bone matrix and overly demineralized bone matrix possessed a degree ofosteoinductive potential whereas bone demineralized to levels of approximately 2%residual calcium provided for maximum osteoinductive potential in both assay sys-tems. The osteoinductive potential of ground demineralized bone varied relative to theparticle size such that DBM particles ranging from 500 to 710 microns provided forthe highest level of calcium deposition (increase of 8.1 weight percent calcium) after4 weeks of implantation in muscle pouches of an athymic mouse, whereas explantedparticles less than 250 microns showed the lowest level of calcium deposition (increaseof only 2.8 weight percent calcium). In the donor age and gender study, DBM fromdifferent donors were divided into 5 age groups for both female and male donorderived bone: less than 20, 21 to 30, 31 to 40, 41 to 50, and 51 to 60 year old agegroups. This study indicated that DBM from female donors in the 31 to 40 years oldage group and male donors in the 41 to 50 year age group possess the highest os-

teoinductive potential, whereas DBM derived from donor bone from both female andmale donors in the 51 to 60 year age group presented the lowest osteoinductive po-tential. DBM derived from male and female donors did not in general show significantdifferences in osteoinductive potential. J Periodontol 1997;68:1085-1092.

Key Words: Biological assay; bone regeneration; bone matrix; alkaline Phosphatase;Osteogenesis; periodontal diseases/physiopathology.

As early as 1889, Senn1 reported using demineralized bo-vine bone as a vehicle for delivery of antiseptics (iodo-form) in patients with osteomyelitis. In the twentieth cen-

tury Leriche and Policard,2 LaCroix,3 Levander,4 Urist,5and Huggins et al.,6 as pioneers, studied induced boneformation. The first unequivocal demonstration of matrixinduced bone formation was by Urist5 in 1965 in a reportdescribing specific preparations of allogeneic bone matriximplanted in muscle.

Due to its remarkable regenerative ability, bone is one

*Center for Biotechnology, Old Dominion University, Norfolk, VA.fLifeNet, Virginia Beach, VA.

of the most frequently transplanted tissues in humans andis routinely used for the repair of skeletal defects causedby trauma, neoplasia, and infection. Three mechanismsmay contribute to the deposition of bone after bone graft-ing: Osteogenesis, osteoinduction, and osteoconduction.Osteogenesis is the formation of new bone from bone-forming cells (osteoblasts) that are transplanted as a via-ble cellular component in autogenous bone grafts. Os-teoinduction is the formation of new bone by recipientmesenchymal cells that differentiate into bone-formingcells under the stimulation of matrix and associated pro-tein factors present in demineralized bone. Osteoconduc-tion is a process in which host bone-forming cells infil-

1086 EFFECTS OF DEMINERALIZATION ON DBMJ Periodontol

November 1997

trate, proliferate, and form new bone in a favorable en-

vironment, such as autogenous bone scaffold or other ma-

trices such as hydroxyapatite from coral.Osteoinduction associated with implantation of demin-

eralized bone matrix (DBM), also designated as DFDBA,is mainly due to the presence of a proteoglycan/collagenmatrix and protein factors such as the bone morphoge-netic proteins (BMPs).79 These bone morphogenetic pro-teins can be extracted from DBM with guanidinium hy-drochloride;10-12 however, after separation, neither the ex-

tracted proteins (BMPs) nor the residue (extracted bonematrix) is capable of inducing bone formation. These twocomponents of DBM can be recombined to reconstitute a

bioactive bone-inducing matrix.13 The appearance of fi-bronectin on the surface of implanted DBM particles hasbeen demonstrated to be ubiquitous throughout the de-velopment of endochondral bone and bone marrow.14

Previous studies have demonstrated that calcium phos-phates (hydroxyapatite), as normal components of bone,may hinder bone morphogenesis.15 In addition, physicalcharacteristics of DBM such as particle size1617 and othercomplex factors such as the donor age of the implantedgraft and the age of host18 have also been shown to affectthe osteoinductivity of DBM. The present study demon-strates that the demineralization process, bone particlesize, and gender of the donor can impact significantly on

the osteoinductivity of DBM.

MATERIALS AND METHODS

Preparation of Demineralized Bone Matrix (DBM)DBM with variable residual calcium. Ground bone ma-

trix (size from less than 250 to 850 microns) provided bythe manufacturer1 was demineralized by exposure to di-luted hydrochloric acid. Briefly, ground bone matrix was

exposed to 0.5 HCL such that solubilized calciumphosphate salts did not neutralize the acid solution anddemineralized bone matrices of variable calcium contentwere obtained by removing bone matrix from the acid atprescribed time intervals; i.e., 45, 90, 135, and 180 min-utes (note that, using the manufacturer's technology, de-mineralization of ground bone can be accomplished in a

reproducible manner where percent residual calcium inthe bone materials is a linear function of pH of the eluentdemineralization solution), and then following a pro-longed demineralization process (5 hours). The deminer-alization process was monitored by the pH of the demin-eralization solution. The variably demineralized bone ma-

trices were washed, freeze dried, and stored at—

80°C.The demineralization process was performed asepticallywith no additional sterilization of the DBM prior to use

in the bioassays. Residual calcium levels were measured

using a modification of the Arsenazo III calcium assay*as described by Zhang.19

Preparation of DBM with different particle sizes.Bone chips from the manufacturer* were ground by im-pact fragmentation and separated using a series of vari-ably sized sieves. Samples were collected according tothe following particle size ranges: <250 microns, 250 to350 microns, 350 to 550 microns, 550 to 710 microns,and 710 to 850 microns. Samples were then demineral-ized in 0.5 HCl where the demineralization process was

stopped when the pH of the demineralization solutiondropped to pH 1.

In Vivo BioassayTen-week old, male athymic mice§ were anesthetized. Im-plants of rehydrated DBM were placed into musclepouches created bilaterally within the longissimus dorsimuscle. The calcium content of all implanted DBM wasdetermined using the Arsenazo III procedure.19 After 4weeks of implantation, the implants were removed andcleaned of excess tissue. The expiant from each implan-tation site was dried at 95°C, weighed, ashed at 600°Covernight, and dissolved in 1 HCl. Aliquots of eachsample were then mixed with 0.05% Arsenazo III in 0.5MTris-HCl buffer (pH 7.4) and absorbance read at 650 nm.The calcium contents of all samples were calculated usinga standard curve of known concentrations of calcium car-

bonate and expressed as a weight percent of freeze-driedmaterial. This bioassay measures changes in percent re-sidual calcium following 4 weeks of implantation in mus-cle pouches of athymic nude mice; i.e., it is a reminer-alization assay. However, in an effort to accommodatecommon usage, the authors have used the terms "osteoin-ductive" or "osteoinductive potential" when referring tothe results of the in vivo bioassay.

In Vitro BioassayFor the in vitro bioassays, human periosteal (HPO/CB-MZ01) cells20 were removed from the cell bank, thawed,and cultured in alpha-MEM* supplemented with 10% FBSfor a minimum of one passage in order to obtain sufficientnumbers of cells for experimental use. The in vitro bio-assays were conducted as previously described.20 Briefly,human periosteal cells, passage 5, were seeded into T-25flasks at a concentration of 2.5 104 cells/cm2. The cellswere maintained in alpha-MEM supplemented with 10%FBS, penicillin, and streptomycin19 until reaching conflu-ence. The alpha-MEM medium was then changed to Dul-becco's MEMf supplemented with 2% FBS and antibi-otics. Demineralized bone matrix or non-demineralizedbone matrix (as a control) was added to appropriate flasks

•Sigma Chemical Company, St. Louis, MO.sHarlan Laboratory, Indianapolis, IN.

Volume 68Number 11 ZHANG, POWERS, WOLFINBARGER 1087

at 5 mg bone material/flask except where designated inspecific experiments.

Alkaline Phosphatase (ALP) activity was measured bythe method described by Wolfinbarger and Zheng,20 whereHPO cells were washed, harvested, and sonicated. Briefly,one ml aliquots were mixed with 0.2 ml of 0.1 mmole/mlp-nitrophenyl phosphate in 0.15 M 2-amino-2-methyl-l-propanol buffer, pH 10.4, and incubated at 37°C for 15minutes. The reaction was stopped by addition of 50 µ of 1 NaOH and absorbance at 450 nm measured. Pro-tein concentrations of the samples were determined usingBCA protein assay kits11 with bovine serum albumin usedfor generation of standard curves. The ALP activitieswere expressed as units of enzyme (µ p-nitrophenyl/min/mg protein). This in vitro bioassay assesses the abil-ity of biomaterials to "induce" increased levels of theenzyme alkaline Phosphatase in the cultured cells and isthus not actually a measure of osteoinductive potential.The terms "osteoinductivity" or "osteoinductive poten-tial" have, however, been used in expressing the resultsof this bioassay in an effort to convey common usage ofthis type of assay.

Statistical AnalysisA minimum of 3 mice were implanted (2 implants/mouse)or 3 flasks of confluent HPO cells were treated for eachexperimental condition tested. Since each sample testedfor percent residual calcium, or ALP activity, consists ofa minimum of 3 replicate assays, a typical data point plot-ted in a figure consists of a minimum of 9 to 18 replicateassays. In addition, in the donor age/gender studies, eachdata point represents a minimum of 5 donors/age group.Prior to statistical evaluation the data were tested for nor-

mal distribution. Linear regression analyses were used toderive standard curves and analyses of variance (ANO-VA) were used to determine the significance among treat-ment groups. Fisher's multiple tests were used for com-

paring means of more than two groups in one way ANO-VA analyses. Significance level was assigned at < 0.05.

RESULTS

Preparation of DBM With Variable Levels ofResidual CalciumGround bone matrix in the size range 250 to 710 microns,obtained from the manufacturer1, was demineralized in0.5 HCL. Residual calcium contents of 32.7, 21.0, 8.9,2.2, 2.0, and 1.2 weight percent were obtained at demin-eralized times of 0, 45, 90, 135, 180, and 300 minutes(Table 1). The pH of the demineralization solution was

approximately 2.4 at the beginning of the demineraliza-tion process due to the buffering ability of the calciumphosphate salts being solubilized from the bone matrix.

"Pierce Chemical Company, Rockford, IL.

Table 1. Residual Calcium as a Function of Demineralization Time

Time pHResidual Calcium(weight percent)

04590

135180300

2.41.751.041.040.750.65

32.721.0

8.92.22.01.2

e•3 Sd 1 £>B Q1) «

ß M.a . *oo pa

10 j8

-

6 -

4 -

2

0

-2 +-4

32.7

li

21 8.9 2.2 1.2

Residual Calcium At Time Of Implantation (weight percent)Figure 1. Changes in percent residual calcium of explanted DBM con-

taining variable levels of calcium at the time of implantation into musclepouches of athymic mice. Equivalent amounts of bone matrix containingvariable levels of residual calcium as shown on X-axis were implantedin muscle pouches. Samples were randomly assigned to each musclepouch and mouse. After 4 weeks of implantation the implants were ex-

planted and cleaned of muscle and connective tissue. The calcium levelswere determined using the Arsenazo III Ca assay and expressed as

change in weight percent residual calcium. Particle sizes of DBM usedhere were 250 to 710 µm.

As the demineralization process continued, the bufferingability of solubilized salts decreased and by 300 minutesof demineralization the pH of the eluent acid solution haddeclined to approximately pH 0.65. In general, before theresidual calcium content reached 2.2% and eluent pHreached 1.04, the demineralization process was fast andmost of the minerals were solubilized. After this point thedemineralization process was slower with minimalchanges in the residual calcium content and pH values ofthe eluent solution.

In Vivo Bioassay and the Effect(s) of ResidualCalcium on Osteoinductive Potential of DBMThe effects of residual calcium content on the osteoin-ductivities of DBM were assessed using the in vivo bio-assay. The changes in percent residual calcium of eachexplant, as a difference in percent residual calcium at thetime of implantation, are shown in Figure 1. Non-demin-eralized bone matrix did not show any new bone forma-tion, but rather exhibited a decrease in calcium content ofapproximately 2.6%. The implants with 21% residual cal-cium at the time of implantation showed a modest in-crease in calcium content, increasing by an average of4.1%. The implants with 8.9% residual calcium showedan average increase in calcium of 5.5% and the implants

1088 EFFECTS OF DEMINERALIZATION ON DBMJ Periodontol

November 1997

with 2.2% residual calcium showed an average increasein calcium of 6.9%. The implants with 2% residual cal-cium showed the highest osteoinductivity of DBM testedwith the calcium content increasing by an average of7.9%. In general, the relationship of the residual calciumat the time of implantation, over the preimplantation cal-cium concentration range of 32.7% to 2%, and "osteoin-ductivity" (remineralization) of the DBM can be ex-

pressed as: Y = -0.28X + 8.34, (R2 = 0.7694), whereY is the weight percent calcium content of expiant and Xis the residual calcium level of demineralized bone matrixat the time of implantation. Between residual calciumcontents of 32.7 (nondemineralized bone) to 2 weight per-cent, as the residual calcium content decreases the os-

teoinductivity of DBM increases. Prolonged deminerali-zation of DBM to residual calcium levels of 1.2% resultedin lower levels of osteoinductive potential such that cal-cium levels increased by only an average of 5.1%. whenimplanted in nude mice.

ANOVA analysis indicated that DBM with differentdegrees of residual calcium showed significant differencesin osteoinductivity (P > 0.05). Fisher's multiple compar-ison test indicated that non-demineralized bone matrix(32.7% residual calcium) had significantly lower osteoin-ductivity than DBM with residual calcium contents of21.0, 8.9, 2.2, 2, and 1.2%. The 2% and 2.2% residualcalcium groups were not significantly different in osteoin-ductive potential, yet both were significantly more os-teoinductive than DBM in the 32.7, 21.0, 8.9, and 1.2%residual calcium groups. There were no significant dif-ferences in osteoinductivities among the 21.0, 8.9, and1.2% residual calcium groups.

In vitro bioassay of the effect(s) of residual calcium.The in vitro bioassay was used as a second means ofassessing the effect of residual calcium on the osteoin-ductivity of DBM. As shown in Figure 2, non-deminer-alized bone matrix did not result in an increase in ALPactivity as compared to control cultures not receivingbone materials. All demineralized bone matrices, irre-spective of the levels of residual calcium, resulted inchanges in ALP activity in the in vitro bioassay. For sam-

ples of bone matrix containing 21% to 2% residual cal-cium, as the residual calcium levels decreased, the abilityto induce ALP activity increased. The DBM demineral-ized to a residual calcium content of 2% produced thehighest levels of ALP activity among the variable residualcalcium level groups tested (Fig. 2). The demineralizedbone matrix containing 1.2% residual calcium at the timeof addition to cultured HPO cells induced ALP activitywhich approximated 71% of the ALP activity induced bythe 2% residual calcium bone, indicating a reduction inosteoinductive potential. ANOVA analysis indicated thatthere were significant differences among the "osteoin-ductivity" of DBM with different levels of residual cal-cium when tested in the in vitro assay (P > 0.05). Fisher's

«I o2 §2 -C•S o<

'

cu o

ce H

I-aII3^

54.5

43.5

32.5

21.5

10.5

0

ri fi

Residual Calcium Levels In Bone Matrix (weight percent)Figure 2. Alkaline Phosphatase levels in human periosteal cells culturedin the presence of demineralized bone particles containing variable lev-els of residual calcium. Equivalent amounts (5 mg) of DBM with differ-ent levels of residual calcium were put into T-25 flasks containing con-

fluent periosteal cells and changes in ALP activity were assayed on day5 after DBM addition. Particle sizes of DBM used here were 250 to 710pjn.

gf 6-

s . 2 +

J 01 +

rfi +- -+-

rfi-+-

710- 500- 355- 250-850 710 500 355

Bone Particle Size (microns)

<250

Figure 3. The effect ofparticle size on the remineralization of implants,in vivo bioassay. DBM with different particle sizes and 2% residualcalcium content were randomly assigned into muscle pouches and mice.After 4 weeks of implantation the athymic mice were sacrificed, implantswere dissected, and calcium contents were assayed using the ArsenazoHI assay.

multiple comparison test revealed that non-demineralizedbone matrix (32.7% calcium content) and the extensivelydemineralized bone matrix (1.2% calcium content) were

not significantly different in induced ALP activities thancontrol cells; i.e., cells not receiving bone matrix. All oth-er demineralized bone matrix groups had significantlyhigher ALP levels than control cells and bone matrix with2% residual calcium resulted in the highest ALP levels inthe cultured human periosteal cells.

In vivo bioassay and the effects ofparticle size. Theeffect(s) of particle size on osteoinductivity of DBM were

assessed using the in vivo bioassay. All DBM used in thisparticle size study contained approximately 2% residualcalcium. The effects of particle size differences on os-

teoinductive potential in the in vivo bioassay are shownin Figure 3. DBM in the 500 to 710 microns size rangeshowed the highest osteoinductive potential, where cal-

Volume 68Number 11 ZHANG, POWERS, WOLFINBARGER 1089

18 T16 +-50 14§12 +li 10 i

a.

64 +20

710-850

500-710

355-500

250-355

<250

Bone Particle Size (micron)Figure 4. Expiant weights at the time of explantation as a function ofbone particle size range at the time of implantation.

cium content increased by 6.7% after 4 weeks of implan-tation in muscle pouches in nude mice. From the 500 to710 microns particle size range DBM to the <250 mi-crons particle size range of DBM, as the particle sizedecreased the new bone formation induced by the DBMdecreased. DBM bone particles in the 350 to 550 micronsand 250 to 350 microns size groups resulted in equivalentlevels of new bone formation (increasing in calcium con-

tents by 3.7% and 4.0%, respectively), and DBM in the<250 microns particle size group provided for increasesin calcium content of only 2%. In general, for groundbone demineralized to approximately 2% residual calciumusing the manufacturer's proprietary technology, DBM inthe 550 to 710 microns particle size were maximally os-teoinductive when compared to similarly demineralizedDBM of different particle size ranges. Smaller particlesizes and large particle sizes provided for lower levels ofosteoinductivity in the in vivo bioassay. Weight changes(Fig. 4) in explanted materials over the 4-week implan-tation showed that expiants from the 250 to 350 micronsparticle size group and the <250 microns particle sizegroup showed more weight loss than expiants containingthe larger bone particle sizes. ANOVA analysis indicatedthat there were significant differences among the osteoin-ductivities of DBM in the different bone particle sizegroups (P < 0.05). Fisher's multiple comparison test in-dicated that the 550 to 710 microns bone particle sizegroup had significantly higher osteoinductive potentialthan DBM in all other bone particle size groups tested,the <250 microns bone particle size group had signifi-cantly lower osteoinductive potential than all other boneparticle size groups tested, and there were no significantdifferences in osteoinductive potential among the DBMin the 350 to 550 microns, 250 to 350 microns, and 710to 850 microns bone particle size groups.

In vitro bioassay of the ejfect(s) ofparticle size. TheALP activities induced by different particle sizes of DBMand control cultures (cells receiving nondemineralizedground bone) are illustrated in Figure 5. All DBM used

Sco G

8*3M Sc eviI

G,

4.5 T

4 -

3.5 -

3 -

2.5--2--1.5--1- 0.5--0--

coo

, omom"!<n

ir¡O

es

ov>o)V

Bone Particle Size (microns)Figure 5. Alkaline Phosphatase levels in human periosteal cells incu-bated in the presence of ground demineralized bone particles (with 2 residual calcium level) of variable particle size ranges. Equivalentamounts of different particle size ranges of DBM (5 mg) were added to

confluent cultures of human periosteal cells. Control flasks consisted ofconfluent cultures which did not receive DBM. These cell cultures were

incubated for 5 days and the induced ALP activities assayed on day 5.

in these studies of particle size effects contained approx-imately 2% residual calcium. DBM in the 550 to 710microns bone particle size range induced the highest lev-els of alkaline Phosphatase in the confluent cultures ofHPO cells. As bone particle size increased or decreasedaround the 550 to 710 microns bone particle size range,the induced ALP levels decreased, such that the 350 to550 microns and 250 to 350 microns bone particle sizegroups provided for roughtly equivalent levels of inducedALP activities. DBM in the <250 microns bone particlesize range provided for induced ALP activities approxi-mating 43% of the highest ALP activities observed, andDBM in the 710 to 850 microns bone particle size rangegroup provided for ALP activities which approximated50% of the highest ALP activities observed. ANOVAanalysis indicated that there were significant differencesamong the induced ALP activities in the HPO cells treat-ed with DBM in the different bone particle size rangestested (P > 0.05). Fisher's multiple comparison test in-dicated that all the size groups tested, except for the <250microns bone particle size group, induced significantlyhigher ALP activities than the control cells (cells receiv-ing nondemineralized ground bone). There were no sig-nificant differences between ALP activities induced byDBM in the bone particle size group 710 to 850 micronsand the <250 microns bone particle size group. Therewere also no significant differences in induced ALP levelsby DBM in the 350 to 550 microns and the 250 to 350microns bone particle size groups. DBM in the 550 to710 microns bone particle size group provided for signif-icantly higher levels of induced ALP activity than DBMin all other size groups tested.

In vivo bioassays of donor-related effects on osteoin-ductivity ofDBM. Because donor age and gender are pre-sumed to affect the osteoinductive potential of derivedDBM, similarly demineralized bone from variable age

1090 EFFECTS OF DEMINERALIZATION ON DBMJ Periodontol

November 1997

, jsa 00

8 "g

1614

--

12--

10

« S gO 'So

4

2 +0

m

<20 21-30

31-40

Age Group

41-50

fa

51-60

Females Male

Figure 6. The effect of donor age and gender on the osteoinductivity ofDBM. DBM from different donors was grouped into the age and gendergroups as described and put into muscle pouches for 4 weeks. Osteoin-duction of expiants was measured using the Arsenazo III assay and ex-

pressed as weight percent calcium of expiant dry weight. Data representthe mean ±SD (n=5). DBM used here contained 2% residual calciumand size ranged from 250 to 710 µ/ .

groups of both male and female donors were implantedin muscle pouches of athymic mice. All DBM used inthis gender and age effects study contained approximately2% residual calcium and contained bone particles in thesize range from 250 to 710 microns. The implanted ma-

terials were explanted after 4 weeks and processed foranalysis of residual calcium. As illustrated in Figure 6,there were both donor age and gender-related effects on

the osteoinductive potential, as measured in the athymicmouse model, of ground demineralized bone (DBM).DBM derived from female donors appears to be maxi-mally and significantly more osteoinductive (P < 0.05)when derived from donors in the age group 31 to 40 yearsof age than from donors in all other age groups tested.DBM derived from male donors appears to be maximallyand significantly more osteoinductive (P < 0.05) whenderived from donors in the age group 41 to 50 years ofage than from donors in all other age groups tested. DBMderived from all age groups of both male and female do-nors were remineralized when implanted in this mouse

model system and thus it may be suggested that DBMcan be prepared from a wide age range of donors and thatbone from older male and female donors may be expectedto stimulate new bone formation when implanted in clin-ical applications.DISCUSSIONOsteoinductivity of DBM was assessed by measuringchanges in weight percent residual calcium following im-plantation in muscle pouches in the nude mouse and bymeasuring induction of cell differentiation (increased lev-els of ALP activity) following addition of demineralizedbone to cultures of confluent human periosteal cells. Itwas demonstrated that the percent residual calcium couldbe used to predict optimal levels of demineralization, such

that bone which was under- or over-demineralized was

clearly less osteoinductive. Available data suggest thatground bone demineralized to an approximate level of 2%residual calcium is optimally osteoinductive resulting innew bone formation. It is thus essential that the deminer-alization process be monitored to ensure that ground boneis reproducibly demineralized. It should be emphasizedthat the authors are not suggesting that percent residualcalcium in DBM is the sole determinant, or even the mostimportant determinant, of osteoinductive potential ofDMB. Rather, it is suggested that the percent residual cal-cium may be used as one indicator of when the deminer-alization process has progressed to a point where theDBM will be maximally osteoinductive, as defined by thecurrent in vivo and in vitro assays. The mechanism(s) bywhich DBM stimulates new bone formation (as a com-

plex sequence of events) presumably involves residualcalcium levels (perhaps as nucleation sites for the rede-position of calcium phosphates), degradation of the or-

ganic (collagen/proteoglycan) matrix (as elements whichmay diffuse from the implant stimulating cellular Che-motaxis into the implant and/or as sites to which cellsattach and receive appropriate regulatory signals), releaseof growth factors (which may not only stimulate cell in-filtration, but induce differentiation of mesenchymal cellsinto matrix and bone-forming cells), and preparation of a

matrix which facilitates the infiltration and attachment ofthe right type of cells in some proper, but as yet unknown,sequence.

The results of this study revealed that non-demineral-ized bone matrix is not osteoinductive; that slightly de-mineralized bone matrix provided a degree of osteoin-ductive potential, as the mineral content decreased theosteoinductivity increased; and that overly demineralizedbone matrix provided for reduced osteoinductivity. Uristand Strates'5 reported that low yields of new bone were

produced in implants of partially demineralized bone ma-

trix preparations and high yields of new bone were ob-tained in implants of nearly or completely demineralizedmatrix. They suggested that old bone mineral delays theprocess of reabsorption and reduces the yields of new

bone. Based on information presented in this study, it maybe suggested that old bone mineral does indeed delayreabsorption; however, provided that there is not an ex-

cess of old bone mineral, the presence of some old bonemineral enhances subsequent bone induction by implant-ed demineralized bone.

The BMP hypothesis proposed by Urist21 suggestedthat BMPs are released from a supra-molecular aggregateof non-collagenous proteins in the process of normal boneturnover or in response to injury or transplantation. Theobservation in this study; i.e., that the non-demineralizedbone matrix showed no osteoinductivity, is presumablydue to the generally accepted notion that BMPs are

trapped by the minerals, and demineralization will release

Volume 68Number 11 ZHANG, POWERS, WOLFINBARGER 1091

BMPs roughly proportional to the extent of deminerali-zation. Presumably, as more minerals are removed, more

BMPs become exposed and solubilizable from the DBMwith the result that properly demineralized bone is max-

imally osteoinductive. Since remineralization of bone re-

quires an initial nucleation event(s), it may be suggestedthat, by leaving some amounts of residual calcium (ascalcium phosphates) in demineralized bone, these foci ofhydroxyapatite facilitate calcium deposition by providingsites for deposition of calcium phosphate salts. Over-de-mineralization may reduce (or virtually eliminate) thesefoci and remineralization would then necessitate nucle-ation events which would presumably require higher con-

centrations of calcium and phosphate than required forsimple crystal growth. Consequently,' demineralization ofbone must represent a compromise in removal of enoughmineral to facilitate release of optimal levels of solublefactors, yet leave ßµ ß amounts of crystalline calciumphosphate to facilitate redeposition of mineral in the re-

modeled and recellularized matrix.Additional data presented in this study demonstrate that

bone particle size has a dramatic effect on the osteoin-duction process. Bone particles between 550 microns and710 microns appear to represent the optimal size range inthat bone particles of larger or smaller sizes remineralizedto a lesser extent in the in vivo assays and stimulatedlower alkaline Phosphatase activities in the in vitro as-

says. Similar results have been reported by Vail et al.16who found that equine bone matrix particle sizes of 2.0mm3 and 5.0 mm3 were associated with osteoinductiveactivity and minimal signs of local inflammation. The twosmaller particle sizes (0.425 mm3 to 0.850 mm3 and 0.85mm3 to 2.0 mm3) were minimally osteoinductive and were

associated with a greater local inflammatory response. Forthe present size study, different particle sizes of DBMwere demineralized for equivalent periods of time tomimic the actual process of demineralizing bone in the250 micron to 710 micron particle size range. The resid-ual calcium results indicated that different particle-sizedDBM contained different levels of residual calcium wheresmaller particle-sized DBM was more easily demineral-ized than larger particle-sized DBM. Further experimentswith different particle-sized DBM demineralized to iden-tical residual calcium levels will need to be performed todetermine the relationship(s) between bone particle sizeand percent residual calcium as it pertains to the potentialfor stimulation of new bone formation.

Additional factors must be considered in assessing therespective roles of bone particle size in both the in vivoand in vitro assay. It was demonstrated in the presentstudy that demineralization of ground bone containing a

range of particle sizes results in variable levels of demin-eralization. Smaller bone particles are more extensivelydemineralized than larger bone particles within a giventime interval. Thus, within a specific sample of ground

demineralized bone, the percent residual calcium repre-sents an average level of demineralization and the com-

posite product represents a complex interplay of residualcalcium and particle size when used in some clinical ap-plication. Future investigations will need to consider therole of both particle size and percent residual calcium. Itis emphasized that the data in the current study regardingpercent residual calcium and subsequent osteoinductivityestimates should not be construed to suggest that percentresidual calcium is the sole determinant of osteoinductiv-ity. Rather, the demineralization process must be consid-ered to impact not only on percent calcium but on themolecular and structural properties of the residual proteincomponents, including collagenous and noncollagenousproteins, and this demineralization event results in a com-

plex biomaterial which dictates subsequent cellular andhumoral responses when used in clinical applications.

AcknowledgmentsThis research was supported by a grant from LifeNet. Dr.Powers is an employee of LifeNet. Dr. Wolfinbarger is a

consultant to LifeNet and currently serves in a honoraryposition as the Director of Research and Development forLifeNet.

REFERENCES1. Senn N. On the healing of aseptic bone cavities by implantation of

antiseptic decalcified bone. Am J Med Sci 1889;98:219.2. Leriche R, Policard A. The Normal and Pathological Physiology of

Bone: Its Problems (Translated from the French by SherwoodMoore and JA Key). St. Louis: CV Mosby Company; 1928.

3. LaCroix P. Organizers and growth of bone. J Bone Joint Surg 1947;29:292-296.

4. Levander G. Induction Phenomena in Tissue Regeneration. Balti-more: Williams and Wilkins Company; 1964.

5. Urist MR. Bone: Formation by autoinduction. Science 1965;150:893-899.

6. Huggins C, Wiseman S, Reddi AH. Transformation of fibroblasts byallogeneic and xenogeneic transplants of demineralized tooth andbone. J Exp Med 1970;132:1250-1266.

7. Urist MR. New bone formation induced in postfetal life by bonemorphogenic protein. In: Becker RO, ed. Mechanisms of GrowthControl. Springfield, IL: Charles Thomas; 1981:406-434.

8. Urist MR, Lietze A, Mizutani H, et al. A bovine low molecularweight bone morphogenetic protein (BMP) fraction. Clin Orthop1982;162:219-232.

9. Urist MR, Sato K, Brownell AG, et al. Human bone morphogeneticprotein (hBMP). Proc Soc Exper Bio Med 1983;173:194-199.

10. Anastassiades T, Puzic O, Puzic R. Effect of solubilized bone matrixcomponents on cultured fibroblasts derived from neonatal rat tissues.Calcif Tissue Res 1978;26:173-179.

11. Takaoka K, Ono K, Amitani K, Kishimoto R, Nakata Y. Solubili-zation and concentration of a bone inducing substance from a murineosteosarcoma. Clin Orthop 1980;148:274-280.

12. Dubuc FL, Urist MR. The accessibility of the bone induction prin-ciple in surface decalcified bone implants. Clin Orthop 1967;55:217.

13. Sampath TK, Reddi AH. Homology of bone-inductive proteins fromhuman, monkey, bovine, and rat extracellular matrix. Proc NatAcadSci 1983;80:6591-6595.

14. Weiss RE, Reddi AH. Appearance of fibronectin during the differ-

1092 EFFECTS OF DEMINERALIZATION ON DBMJ Periodontol

November 1997

entiation of cartilage, bone, and bone marrow. J Cell Biol 1981;88:630-636.

15. Urist MR, Strates BS. Bone formation in implants of partially andwholly demineralized bone matrix. Clin Orthop 1970;71: 271-278.

16. Vail TB, Trotter GW, Powers BE. Equine demineralized bone ma-

trix: relationship between particle size and osteoinduction. Vet Surg1994;23(5):386-395.

17. Syftestad G, Urist MR. Degradation of bone morphogenetic activityby pulverization. Clin Orthop 1979;141:281-286.

18. Syftestad G, Urist MR. Bone aging. Clin Orthop 1982;162:288-297.19. Zhang M. The assessment of osteoinductivity in human allograft

demineralized bone matrix by in vivo and in vitro assay models.

[Thesis] Norfolk, Virginia: Old Dominion University and EasternVirginia Medical School; 1996. 179 p.

20. Wolfinbarger L, Zheng Y. An in vitro bioassay to assess biologicalactivity in demineralized bone. In Vitro Cell Dev Biol Anim 1993;29A:914-916.

21. Urist MR. Bone morphogenetic protein. In: Habal MB, Reddi AH,eds. Bone Grafts & Bone Substitutes. Philadelphia: W.B. SaundersCompany; 1992:70-83.

Send reprint requests to: Dr. Lloyd Wolfinbarger, Jr., LifeNet, 5809Ward Court, Virginia Beach, VA 23455.

Accepted for publication March 24, 1997.