You are on page 1of 7


Full=ArchImplant Framework Casting
Accuracv: Preliminam In Vitro
Observation for In Vivo Testing
Alan Brooks Carr, DMD, MS, * and Robert Bruce Stewart, DDS,MS#
Purpose: Conventional techniques for implant metal framework fabrication produce error of a
magnitude that is inconsistent with the passive-fit requirement for osseointegrated implants. To
understand the correlation between prosthesis fit and the implant-tissue response, evaluation of
the interface tissue reactions t o customary levels of fit is required. The purpose of this study is t o
determine the accuracy of torch casting full arch frameworks using a high palladium alloy and a
ringless phosphate-bonded investment technique.
Materials and Methods: Three different variables were considered relative t o casting accuracy
effect. The first variable, completeness of mold-fill, compared cast specimens where the entire
sprue system was filled as part of the casting and cast specimens without the sprue system filled.
The second variable, phosphate-bonded investment special liquid concentrations, compared
groups of castings produced from 0%. 12%. 25%. and 50% special liquid. The third variable,
investment mold shape, compared casting produced from a conventional ringless mold shape with
a modified ringless mold shape where the investment in the same horizontal plane as the pattern
was equal in thickness at the internal and external surfaces. Horizontal and vertical distances on the
wax pattern and resulting framework were measured using a machinists microscope t o determine
casting error. Combined vertical and horizontal error was used for comparison between groups
(one-way analysis of variance).
Results: No significant differences existed among the three groups compared (P >0.05). The
mean error comparison between the complete and incomplete mold-fill groups showed no
statistical difference, while the incomplete fill group was found to be more porous. The mean error
of all groups (0.130 mm] exceeded the recommended level of fit needed t o satisfy the passive fit
requirement by more than 10-fold.
Conclusions: These results verify clinical observation and suggest that the use of conventional
lost wax casting technique t o cast one-piece full arch implant frameworks is both imprecise and
inaccurate as judged against the passive fit requirement. The consequences of screw-fastening
misfitting prostheses t o osseointegrated implants is currently under investigation.
J Prosthod 2:2-8. Copyright o 1993 by the American College of Prosthodontists.

INDEX WORDS: casting accuracy, implant prostheses, implant-tissue stress


URRENT EKDOSSEOUS implant success has

resulted from basic principles of implant material preparation and design, atraumatic surgical
technique, provision of an unloaded healing phase,
and proper prosthetic fit and loading.' The term
passive fit characterizes the connection between the
cast framework and abutment or implant, and was
described by Brsnemark to exist at the 0.010 mm
*A.ui.stant Prafisor, Directm of Madlofacial Prosthetics, Section of
Ratoratiue und Prosthetic Dentwty, Collqe d-Dentisty, The Ohio State
Uniueni@, Columbus.
~Pricalepractice,Dttruit, MI.
Address repnnt requests to Alan Brwks Can; D M D , IMS, Collegp oJ
Dentisty, Ohw State Universip? 30.5 MJ 12th AL'e, Columbus, OH
Copyright Q 1993 tp t h e i l r n ~ c a nCollege ofPmthodontists
10.59-941Xl93/0201-0001$5.00/ 0


level.*This level of accuracy is a reflection of the rigid
support provided b) osseointegrated implants and
underscores the requirement to achieve optimum
load di~tribution.3,~
Accurate connections sew? to
minimize prestressed loads (ie. loads inherent in the
connecting process) and favorably distribute all external loads, thus reducing unfavorable mechanical and
biological sequelae. Such connections allow an adequate stimulus for remodeling during the stage of
implant-tissue interface maturation toward steadystate.'

Implant prosthesis fabrication technology, borrowed
from conventional prosthodontics, has not been shown
to predictably produce prosthesis fit at the 0.010 inm

Jormal oJProrthodonticJ, Val 2, X o 1 (March), I993:pp 2-8

implant-support ed prost he sis fabrication protocols in use are apparently effective for various clinical situations. A separate split mold was used to fabricate the wax reservoir bar and sprue system. as described later. Pattern dimensions were determincd by pilot measurement to be stable 60 minutes following connection of pattern parts. and vertical measurement points were established by creating depressions on the cast and on thc pattern lateral to the horizontal points.010 mm level.043 mm to 0. To understand the correlation of prosthesis fit and interface response.196 mm. evaluation of interface reactions to customary levels of prosthesis fit is rcquired. judge as acceptable margins exhibiting a horizontal range of opening from 0. Romulus. special liquid concentration.^ Despite the shortcomings associated with fabrication accuracy. where mold-fill. MI). the fixtures.100 mm using conventional techniques for multiunit tooth-supported prosFurthermore. 'The wax pattern and sprue system was placed in the ring so that the runner bar was at the heat center of the investnient block and the framework arch form was 14 mm from the investment periphery and 12 mm from the end (Fig 2). and precast dimensions were recorded twice following the measurement protocol described below.I0-"Although clinical protocols differ in the methods used to control the fit of full-arch prostheses. Although failure of integratcd implants following prosthesis connection is not a frequent complication. it is common that one-piece frameworks are used without efforts to modify their as-cast accuracy. Nobelpharma USA. The purpose of this study was to determine discrepancies in both the horizontal and vertical 3 planes at framework locations simulating terminal abutmcnt positions in a mandibular five-implant arrangement for one-piece castings.The horizontal measurement points were established by creating depressions in the pattern. it is both poorly understood and common enough to warrant investigation of possible prosthodontic cause^. Casting accuracy studies have shown error in the range of 0. IL) was fastened at a midline abutment analog. Figure 1. iVumber I level of accuracy. as exemplified by a recent report describing a significant force introduced when connecting a prosthesis that had an acceptable fit. Materials and Methods The experimental model simulated a five-implant mandibular framework design (Fig 1). using tactile discrimination methods. it has not been shown that implant framework accuracy offit can be detected at the 0. Such depression allowed reproducible positioning of the reflective spheres used to determine distances for the wax patterns and the castings. The single fastening point allowed the most predictable precasting and postcasting connection and subsequent detection of dimensional changes. A gypsum master cast was constructed to allow one horizontal and two vertical measurements at the terminal abutment locations. and the bone. type C.7It has been shown that experienced dentists. Chicago.March 199. Experimental rnodrl with fastened framework showing (A) horizontal sphere positions and (B) vertical sphere positions.230 mm and a vertical range of opening from 0.7.$ Skalak has described biomechanical concerns relative to connection of nonpassive frameworks that include stress in the prosthesis. 4 single screw for a standard gold cylinder (DCA 072.3 The suggested biological response to excessive and undetected stress is implant interface failure. Volume 2.032 mm to 0. Patterns containing the gold cylinders were formed by a split mold filled with heated wax (Blue inlay casting wax. . and investment design were independent variables. Sybron/Kcrr.

and filled to predetermined level in air Bench set for 60 min and stored in sealed plastic bag (100% humidity) CF and ICF Special liquid concentrations: 0: 100. gas (6 psi) and oxygen (13 psi) Broken arm centrifugal casting machine IS 85 alloy fused 2 dwts at a time. burnout. Right top view illustrates spacer design. rate. Nobelpharma ‘gold’cylinder-DCA 072 Full arch. KY) using a plastic investing ring (Clear-Eze. and air abraded with cylinder protection . investment. Burnout. ICF: 24 dwt (no reservoir) Bench cooled. Pattern. four castings for each group. Sullivan.50:50 Investment shape: Spacer and conventional Spacer. 15”F/min rate. positioned in ring. devested. The wax patterns were invested in fine-grain. batch no. two groups of castings. Whip Mix Corp. represented by shaded area. 15”F/rnin. were produced to determine the effect of completely filling the sprue system on accura- Table 1. arid casting protocol followed in this study. and Casting Protocol Pattern Material Shape Spruce Investment Material Quantity Met hod Groups Burnout Oven Method Time Casting Material Groups Cooling Devesting Blue inlay casting wax hard type/class I. 10731 1100 200 gm powder: 48 mL liquid Slow speed mix (250-550 rpm) under vacuum for 90 s. Scottsdale. 1500°F for 60 min 3 h 15 min Ceramic crucible Multiorifice torch using liquid petroleum. placed by brush to critical areas. 5 mm in diameter attached at 4 locations Cera-Pina. which provided a uniform investment coating to the pattern within the horizontal plane. AZ). Louisville. 12:88. Investment.4 Full-Arch Implant Framework Cmtinp Caw and Stmiart 1: n nm 6mrn - II 6mm IOyn Figure 2. M. carbon-free phosphate-bonded investment (Cera-Fina.In one comparison. only CF data used for 25:75 cgroup.25:75. 800°F for 30 min. Table I outlines the pattern. total fusing time < 4 rnin CF: 32 dwt. phosphate-bonded. Accutherni 11-1000 3 rings placed in a cold oven. 7 mm wide at cantilever end Reservoir ‘runner bar’ 7 mm in diameter. Lshape cross section. Schematic illustration of invested pattern dimensions and position. 20 mm cantilever. 6 mm in height. 9 mm wide at the cylinder. 2 mm connecting sprues. CF and ICF groups all used 25:75 special liquid/water ratio For ratio comparisons.

.436 5. and 50:50 (three framcworks).204 26.829 1.332 15. The measurement scheme used 1. Opto-Metric Tools.003 15.789 15. Williams Division/Ivoclar North America.831 15. New York.252 26. as detcrmined by 10 measurements o f a single horizontal distance.5.708 15. Results Three castings were considered miscasts and were not measured. not analyzing for individual effects. All castings were produced by one operator. nu). J.480 15.092 26.219 26. Amherst.220 26.I3 The completely filled (CF) group used sufficient alloy to fill the sprue system and the incompletely filled (ICF) g o u p used only enough alloy to fill the framework pattern mold space.075 26.007 mm.582 5.003 mm and precision of ?0. using only new alloy.187 26. All invested patterns were allowed to set more than 1 hour (generally overnight) and stored in a humid environment. 'The CE' group of castings required 32 dwt to fill the mold spacc and the ICF group required 24 dwt.687 15.769 15. but was sufficiently precise for the measurements made.5 26.707 15. which attempted to distribute equal investment material external and internal to the pattern and casting. Measurement Data for Wax and As-Cast Metal Groups in Millimeters cvm M€X4T> Group Hmizonlal Vertical Ho~iyintal Vertical *CF 26. three frameworks were cast to investigate a unique investment block design (Fig 2.001 mm.089 26.546 15. For statistical analysis. The coordinates were converted to distances by use of the Pythagorean theorem and two measurements were madc by one operator for each distance to be determined. .167 26.636 CE' CF CF ICF ICF ICF ICF Special 1iquid:waterratic 0:100 26.036 26. in the horizontal plane. Measurements were made using a machinist's traveling microscope (Leitz Model UMW.128 16. The alloy was melted in a ceramic crucible using a conventional multiorifice gas/oxygen torch. NY) with electronic digital micrometer heads (Digirnatic Head 164 Series. Four castings were produced for the complete fill and incomplete fill groups.198 16." Data were collected as three point coordinatcs for the horizontal and vertical measurements on each specimen.212 26.231 5.641 26.115 26.668 26. Vertical rneasurcments (R &L) combined.819 5.5.173 26. and cast using a standard broken-arm centrifugal casting machine (Kerr/ Sybron). the effect of altering special 1iquid:waterratios on accuracywas investigated using spccial liquid concentrations of 0: 100 (two frameworks). the horizontal and vertical errors (measurement differences between pattern and casting) were considered as the same type of error and combined.411 castings were made with a high-palladium alloy marketed for implant-supported prostheses (IS 85.21 1 2629. Measurements were repeated as for the wax patterns. ?Unable to collect vertical data. Jelenko.746 1. The latter group attempted to address the possibility that the spr-ue system may have a detrimental effect on the resulting casting accuracy.052 15. The burnout protocol rollowed manufxturers recommendations and was identical for all casting groups (Table 1). An overhead light source was mountcd above the microscope in a fixed position to provide a reflective reference crosshair in the spheres for spatial positioning and measurr- ment.487 15. Japan) having a manufacturers reported accuracy of 20. thcn placed in a cold oven for burnout (Accutherm 11-1000. Armonk.419 25.022 26.57 mm stainlcss steel spheres placed in the depressions at the horizontal and vertical positions as previously described and illustrated in Fig I. 12238 (two Cameworks).532 12:88 50:50 26.921 15.668 15. For this investigation thc measured precision was 0.135 26. 25:75 (four frameworks-CF group).024 26. All castings were bench cooled to room temperature. In another comparison. The bilateral vertical error was comb i n d as one vertical error.450 15.226 26. One horizontal and two vertical distances were determined for each specimen. These include wax pattern and as-cast measurement data for the horizontal and vertical measurements for the six Table 2. two castings for the 0:100 arid 12:88investment groups.275 26. right shaded area).747 Investment Spacer 26.102 26. Casting cylindcrs were inspected for casting imperfections to assure optimal reseating when screw fastening to the abutment. and three castings each for the 50:50 investment and internal spacer investment group. The original measurements and measurement differences are presented in Tables 2 and 3.082 25.F.580 16.738 5.199 ~~~ *CF group scrvrs as 2575 spccial1iquid:water group. Volum 2.159 26.318 26. This amount of imprecision was caused by the variable roughness effect of the cast framework depressions on the sphere position.149 26. Any discrepancy affecting screw fastening or measurement depressions was considered a miscast and not used in the analysis.241 26.815 15.466 t t 15. therefore.047 26.199 26.5 March 1993. The castings were air-abraded with 50 p n alumina and inspected for casting discrepancies using a 7-30 power stereo microscopc.444 16. by incorporating a spacer within the arch form during investing.448 15. NY). Tokyo.04-6 26. In a third comparison. Wax pattern and as-cast group differences were analyzed statistically for significant differences with the use of one-way analysis ofvariance (AYOVA). Number I q.577 15.241 15. Mitutoyo Mfg.777 15.592 15.730 15.

122 .071 * .0092 0.079 .102 .OM) .0037 0.207 -.05). The vertical difference values showed no consistent trends or correlation to the horizontal values.3 1 . ANOVA Table Group Snurce Sum $ Squares Mean Square F Ratio Probabilip <F Complete fill c incomplete Model Error C Total Model Error C Total Model Error C Total 1 14 15 1 12 13 3 17 20 0.I42 -.058) .I53 .062) .161 mm) and precision (pooled standard dekiation: 0. - - - - . and references each to the entirr experimental population mean (0.1 16 (. because of the qualitative and quantitative difference between the sup- Table 4.0092 0.00.6710 0.2004 - 0.069 .127 .096 (.130 (. Table 4 shows that no significant difference existed among the three comparisons (AKOVA. the horizontal differences showed casting expansion except €or the 0% special liquid and spacer groups.108 .P > .161 (.9157 - 0.1 13 m m u 0.183 .0 126 0.0807 0.0001 0.139 (.127) .119) .I36 .087) . Specifically.090 (.043) .132) .024 -.130 mm) and to zero.. In considering the nonpooled data relative to the expansion and contraction.I16 .5797 - - Complete fill z: spacer Special liquidwater ratios Note: CF also sewes as 25% special liquid group.080 (.142 .069 . Discussion Current implant prosthodontic procedures make use of conventional techniques and materials common to conventional fixed and removable prosthodontics. the clinical significance of such comparisons requires testing in vivo because of the contrasting support models of teeth and implants.166 (. groups of castings.152 (. Although the CF and ICF groups were not significantly different from each other (Fig 3).51 1.08 1 249 .0042 0.089) ..165(.0442 0.0062 0.I52 (.058 (.046 .205 (. The C F group also served as the 25% special liquid group.028) .I37 .I46 . the absolute magnitude of the error was used to provide a one-sided comparison.lo3 .138 (.049 m m u 0.8059 0.089 mm).337 .007 50:50 Internal Investment Spacer .0001 0.0042 0.115 -.1 13 (.1058 0.053) . Figure 3 shows the grouped data.1 184 0.6 Full-Arch Implant Frameamrk Gating Caw and Stewart Table 3.07 14 0.136 . While it is important to compare the accuracy of implant prosthodontic techniques with the accuracy of conventional prosthodontic techniques.070) ..225) .156 (.0117 - 0.361 CF ICF 0:00 1298 . the descriptive data suggcst that the C F group tended to provide better accuracy (pooled mean: 0.098 *Unable to collect vertical data. The ICF group showed greater porosity and no improvement in accuracy.123 (.165 .238 Horizontal Vertical Pooled .074) .lo9 (.I17 .049) .030) .I63 . In the statistical analysis. Differences for Measurement Data in h*lillimeters Absolute Mean fiaerence (SD) Ozflereize Hmzontal Vertical .047) .169 .022 -.071) .

realizing that it may be an oversimplification. The 0% group showed contraction in three of the four measurements (Table 3). porting tissues of teeth and implants. Placement of the reservoirs in the vertical heat center of the investment mold allows for the reservoir to solidify last and also ensures optimum density in the casting. * Figure 3. all such stress is seen as strain by the cell.16 In conventional prosthodontic marginal-discrepancy investigations. causing distortion. If the reservoir were to solidify first. the data are inaccurate and imprecise as judged against the 0. The complete fill group represented the 25% special liquid group. the single fastening point at the anterior abutment positioned each specimen on the cast for measurement. Under these conditions.15and results in stress within the components and at the implant-tissue interfaces of the connected implants if any misfit relationship exists. The internal investment spacer group (25% special liquid) was designed to offset the external expansion ~ . The first part of the study examined the effect of the sprue system on ~ ~ distortion of the framework during solidification. The two best special liquid groups. Screw fastening is a sequential. The ICF group represented the 25% special liquid . 12%.130 mm horizontal line. Implant prosthesis connection and the potential for implant-tissue interface stress mandates not only detailed understanding of accuracy levels for prosthesis fabrication but also the biological effects of connection to the host tissue.’3 Patterns of the size required to cast full-arch restorations require large reservoirs to ensure an adequate sourcc of molten alloy for the solidifying framework. For the comparisons made in this study. The two were considered as similar types of error and represent error in two planes that would be expressed as marginal error in conventional marginal “sleeve fit” accuracy studies. we might not expect the implant response to misfit to be the same as the tooth response to misfit? Potentially as important as the difference in support between teeth and implants is the manner with which the fixed partial dentures are connected to their support. Different data would be expected if the sprue system were placed outside the heat center.” It is impossible to isolate the various strains and correlate them with an observed tissue response in vivo.130 mm). the cumulative error was used. Porosity would also be promoted by not providing a molten reservoir source during framework solidifying. and 50%. The 12% group showed expansion of all dimensions measured.0130 mm. the solidifying framework develops rigidity before the reservoir. Expansion of phosphate-bonded investments can be controlled through alteration of the special liquidwater ratio. active clamping process where each unit is independently connected that provides loads across each screw joint (preloads). A ringless investment mold technique was used to allow optimum expansion. the error attributable to casting is cxpresscd as pooled horizontal and vertical data (Table 3 ) . Volume 2. sufficient stress might be placed on the solidifying framework. Under the conditions of this study. Numbm I i I 1 0 >s 0 14 0 12 Frw. Bar graph showing error (average and s t m dard deviation) by group. thereby reversing the solidifying sequence from the current model and promoting porosity.March 1993. 25%.” Skalak” warns that connection of misfitting prostheses produces stress that can easily escape detection and occurs at a location where the resultant effects are not well understood. It is believed that position of the sprue systcm within the investment block is most critical for accuracy related to mold filling.010 mm requirement. with a mean expansion of 0.3. The results of this casting investigation show that for all variations of the casting procedure evaluated. The overall mean error is represented by the 0. Therefore. the marginal error that is measured results from a combination of horizontal and vertical inaccuracy. along with the spacer group. Therefore. 010 Imml 0 08 0. pooling the errors allows better correlation with precedent casting accuracy liter- 7 ature. special liquid proportions were varied to provide four groups: O%.6 From the perspective of the cellular response to stress. showed data below the entire experimental mean (0.’*In this study. incomplete filling of the reservoir increased the porosity in the rrameworks and did not improve casting accuracy. yet still with an average error 10 times the required level of accuracy. for comparison purposes.06 004 oOW CF ICF Spacer o 0% 12% 509.

Int J Oral Maxillofar Imp1 1990. Given the above findings.. Further study is required to determine the usefulness of such an alternative investment block shape. Without such material support (40 oz do!). Int J Oral Maxillofac Impl 1991. Marketing Brochure. Zarb GA: The longitudinal clinical el1ic. Octobcr. J Dent Res 1992.17 C a n and SteLcialt 5. Dec 8.6270-276 I I. Jemt T. Int J Oral Maxillofac Imp1 1990. Inc: Implant arid Spark Erosion Prosthetic Senices.J Pi-osthet Dent 1989.Dedrnon Hw: Disparity in expert opinion on size of acceptable margin openings. et al: The applicability of osseuintegrated oral irnplants in the rehabilitation of partial edentulism: ?\ prospective multicenter study on 550 fixtures. Academy of Osseointegration Aririiial Session. J Den1 Res 1991. Bergman GF: Laser welding oftitaninm frameworks.49:813-818 4. Bolender C. 1985. Dallas.62:662-668 23. MN 3.JUcritKes 1977. Personal communication. White G: Academy of Osseointegration Annual Session Lecture.2:91-100 12.TX.orcastinginone piece.'^-'^ In spite of the limited sample size of this study. Int J Maxillofac Imp1 1991.J Prosthrt Dcnt 1988.4) 18. Acknowledgment T h e authors acknowledge t h e gracious support of t h e Williams Division of Tvoclar N o r t h Amcrica.6413-417 8. et al: In vivo Ioad measurements on ossrointrgrated implants supporting fixed or removable prostheses: A comparative pilot study. hilosby. Jctnt 'I. van Steenberghe D. pp 117-128 2. February 1990 14. MO. Kubin GT. Lockholm U.Carlssori L.rcy of ossroiritrgrated drntal implants: A three-year report. J Prusthet Urrit 1!%5. May 1991 20. thc overall finding of casting error is in agreement with previous conventional fixed prosthodontic casting accuracy research? Frameworks produced at accuracy levels similar to this study cannot be considercd passive.8 Fu'ull-Arch Implant F r a m e m k Casting effect seen in the CF group by allowing inward expansion at the level of the mold space. Voitik AJ: A direct assembly framework method for osseointegrated implant prostheses. CoxJF. Dental Arts Laboratories. 1991 10. Luby hE. BrSrierriark P-I: Biological principles relative to Osseointegrated implants. requires an understanding of the biological consequences of connecting a misfitting prosthesic to oswointegrated implants. p 369 19.54:i70-7K 6. the data from this study are being applied in ongoing animal studies investigating the implanttissue intcrfacc rcsponse to prosthesis misfit. Carr AB: A Comparison of impression techniques for a five-implant mandibular model. Consequently. Ziebert GJ. Int J Oral Maxillofac Imp1 1987. B r h e m a r k P-I. Skalak R: Biomechanical considerations in osseointegrated prostheses. Lecture presentation. J Prosthet Dent 1983. St Louis. Chicago. Gegauff AG. Albrektsson T (eds): TissueIntegrated Prostheses: Osseointegration in Clinical Dentistry.5:272-281 13. this project would not have brrn possible.. 1985. Dhuru \TI. This oKkt was accomplished in 2 of 3 horizontal errors that showed contraction (Table 3 ) . Rosenstiel SF: 'I'he seating of one-pircr arid soldered fixed partial dentures. Quintessence. Rangert B. the nature of the implant-tissue interface.Jorneus L Forces and moments on branemark implants. Zarb GA. Rochester. conventional soldering. References 1.4:2412. Clark RE: Comparative distortion in three-unit fixed protheses joined by lascr welding. Opcr Drnt 1982. Crit Rev Oral Riol Mrd 1991. Boss A.2:83101 17.62:292297 7. Huling JS. Ail data regarding investment tcchniqucs are specific to the ringless method as described. Craig R G Restorative Dental Ivlaterials.Scliimeger BE. McLeod KJ: Dynamic strain history as a dctcrminant of bone rriorphology. As described earlier. Carr AB. 1989. Preload and load-sharing of strain-gaged CP-Ti implant components. Continuing Education Course. and the scret\-rastening method of rigidly connecting a prosthesis to the supporting implants. Vancouver. Jemt T Failures and coniplications in 391 consecutivrly inserted fixed prostheses supported by Branemark Implants in Edentutous Jaws: A study of treatment from the time of prosthesis placement to the first annual chcckup.7:97-101 9.6:448-455 al: Comparison of accuracy of multi unit onc-piccc castings. BC. strategies for improving the span fit ofprosthcses need to be used and include techniques similar to conventional prosthodontics as well as those unique to implant superstructure^.Brunski JB. The Implant Society 1991f211-14 . Use of metal rings may restrict expansion in the horizontal plane and would bc expected to provide differing results.70:430 (abstr 1310.Zarb GA. February 1992 21. K d e r JCI: The concept of osseointegration and Imne matrix expression. Int J Oral Maxillofac Impl 1991. IL. Srllcrs GC: Direct assembly framework for osseointegrated implant prosthesi5.71: i 2 8 (ahstr 106) 16. Stanford CM.56i:128-134 22.