- Email: [email protected]
Effect of cortical thickness and cancellous bone density on the holding strength of internal fixator screws
ELSEVIER
Journal of Orthopaedic Research
Journal of Orthopaedic Research 22 (2004) 1237-1242
www.elsevier.com/locate/orthres
Effect of cortical thickness and cancellous bone density on the holding strength of internal fixator screws J. Seebeck a, J. Goldhahn a, H. Stadele b, P. Messmer ‘, M.M. Morlock ‘, E. Schneider a
A 0 Research Institute. Davos, Clavadelerstrasse, CH-7270 Davos Platz, Switzerland CA N C A S Research Group, University Hospital, Basel, Switzerland Biomechanics Section, Technical University Hamburg-Harburg, Germany
Abstract Internal fixators are a new class of implants designed to preserve the periosteal blood supply of the bone. In contrast to conventional plate fixation in which the screws have spherical heads and are loaded mainly by axial pullout forces, screws in internal fixators are “locked” within the plate and therefore subjected to axial as well as bending loads. In this study the ultimate loads of screws of a commercially available internal fixator system were tested in a pullout (n = 72) and cantilever bending mode (n = 72) in metaphyseal and diaphyseal regions of four pairs of human tibiae with different bone qualities. Cortical thickness and cancellous bone density were determined at the screw insertion sites. Stepwise multiple linear regression revealed that cortical thickness and cancellous density can explain 93% and 98% of the variance of the ultimate load of the screws in an axial pullout and cantilever bending mode. Screws in internal fixators are better suited to transmit shear forces and thereby make better use of the strength potential of bone than screws used in conventional plate fixation: this is especially advantageous when bone strength is reduced, e.g. due to osteoporosis. 0 2004 Published by Elsevier Ltd. on behalf of Orthopaedic Research Society. Keywords: Screw fixation; Internal fixation; Bone morphology; Fracture treatment
Introduction Screws are an essential element in fracture fixation in the metaphyseal and diaphyseal regions of bone. The maximum load that can be transferred between a screw and the bone is in many cases the limiting factor for the success of the osteosynthesis. In the past, the holding strengths of different screws have been investigated by means of axial pullout tests in cortical and cancellous bone specimens as well as synthetic foams [l-3,5,7, 10,18,24]. The testing of screws in the axial direction was derived from their principal mode of operation in conventional plate fixation. The screws are axially preloaded to generate compression between bone and plate. This compression should result in sufficient friction to transfer loads that act orthogonal to the screw axis. In plate fixation the screw is not supposed to take *Corresponding author. Tel.: +41-81-4142441; fax: +41-814142288. E-mail address: [email protected] (E. Schneider).
significant bending loads to avoid shear failure [26]. This agrees with the common understanding on the usage of screws in mechanical engineering. Some years ago the concept of internal fixators with unicortical screws for treatment of long bone fractures was introduced [15,16]. The initial motivation to develop these types of implants was based on the biological advantage of saving the periosteal blood supply of the bone by avoiding compression between plate and bone. In these internal fixators mechanical stability is maintained by rigidly locking the heads of the unicortical screws within the plate. In this configuration the screws are no longer used in the classical way but act mechanically like the pins of an external fixator. As a consequence the screws are now sustaining a pronounced cantilever bending in addition to axial loads 1161. In a previous study using bovine cortical bone specimens we showed that a screw submitted to bending can transmit two times the loads tolerated in axial pullout before failure of the bone [20,21]. However, no information is available regarding the amount of load that
0736-0266/$ - see front matter 0 2004 Published by Elsevier Ltd. on behalf of Orthopaedic Research Society. doi:10.1016/j.orthres.2004.04.001
1238
J. Seebeck et al. I Journal of Orthopaedic Research 22 (2004) 1237-1242
can be transferred in the metaphyseal and diaphyseal regions of human bone when the screw is subjected to bending vs. axial loads. In this context, the cortical thickness and the cancellous bone density are of importance since insufficient screw anchorage is a serious problem in bone that has become mechanically weak as e.g. in osteoporosis. The purpose of this study was to determine the holding strength of screws used in an internal fixator system in both, an axial pullout and a cantilever bending mode at different sites of the human tibia and to thereby evaluate the influence of cortical thickness and bone density on the holding strength of the internal fixator screws.
Materials and methods Specimen preparation
The cortical index (CI), i.e. the ratio between the cortical thickness and the bone radius in the central cross-section of the diaphysis was determined by means of an antero-posterior X-ray in a pool of 10 pairs of fresh frozen human tibiae [ll]. Two pairs (61 years-old female, 70 years-old male) with the highest CI (0.41,0.48) and two pairs (82 yearsold female, 72 years-old male) with the lowest CI (0.19, 0.23) were chosen. These eight bones were scanned in a clinical CT-Scanner (Somatom 4 volume zoom, Siemens, Erlangen, Germany) using a slice thickness and slice distance of 1 mm each, and an in-plane resolution of 0.234 mm (scan diameter 120 mm, picture size 512x 512 pixels, 4 x 1 mm collimation, 120 kV, 100 mA) according to a previously described setup [12]. A phantom (European Forearm Phantom) [I41 was included in the scan of each bone to later relate the absorption values of the cross-sectional CT images with the apparent bone mineral density. The tibiae were divided into 10 sections of equal lengths. Five proximal and the three most distal bone segments were prepared by transversely cutting each tibia by means of a diamond band saw (Exakt, Norderstedt, Germany). Screw insertion
Commercially available titanium screws (Synthes, LCP-Locking Head Screws 05.0 mm, lengths 46 and 26 mm, Mathys Medical Ltd., Bettlach, Switzerland) were used for both pullout and cantilever bending tests. The screws, designed to be used in both cortical and cancellous bone are self-tapping and have a core diameter of 4.4 mm, an outer diameter of 5 mm and a pitch of 1 mm. The heads of the screws are formed as a conical thread which locks the screw within in the corresponding fixator system [9]. For the pullout and cantilever bending tests a mounting block with the same conical thread replaced the fixator. The number of screws that could be placed into each bone segment was limited by the sizes of the segments. Making sure that each screw did not interfere with the others, three screws were placed in the first proximal metaphyseal segment of all bones except for one pair of bones that was of bigger size and allowed a placement of four screws. In the second proximal segment as well as in the last two distal segments just one screw could be tested in all bones. The remaining diaphyseal bone segments allowed a placement of three screws per segment in accordance with the triangular cross-sectional shape of the tibia1 diaphysis. The screws in the first two proximal segments as well as the last distal segment had a total length of 46 mm while in the other segments screws with a length of 26 mm were used to achieve a unicortical insertion as suggested for the use with internal fixators. In accordance with the surgical technique recommended for this type of screw, 4.3 mm pilot holes were drilled with an insertion point at 50% height of the bone segments and with a depth corresponding to the screw insertion length. A linearly guided machine drill was used
such that the drill axes were adjusted parallel to the segment cutting planes to ensure co-linearity of the screw axes with the distraction force under axial pullout and pure transverse loading in the cantilever bending mode. The self-tapping screws were inserted by hand holding the mounting block in place until the conical head of the screw was locked into it. The size of the mounting block only allowed a sequential insertion and testing of each single screw. In order to record the position and direction of the screw axes 4.0 mm stainless steel cylinders were placed into the predrilled holes before screw insertion and macroradiographs of the bone segments with the inserted cylinders were taken in the proximo-distal direction (cabinet X-ray system, model 43855A, Faxitron X-ray Corporation, IL, USA). Pullout and cantilever bending tests
Screws were axially pulled out from one bone of each pair while in the counterpart the screws were loaded in a cantilever bending mode pulling in the direction of the long axis of the bone (Fig. 1). In the cantilever bending mode, the free bending length of the screw between the locked head and the bone determines the bending moment induced into the screw and has an influence on the load bearing at the bonescrew interface. Therefore the free bending length was kept constant by placing a spacer of 5 mm thickness between the mounting block and the bone surface during screw insertion. In addition, cantilever bending tests of screws with free bending lengths of 1 and 3 mm were performed for 20 specimens to determine the effect of the free bending length. In both loading modes, the bone segments were rigidly connected to the base plate. The midpoint of the screw heads were aligned with the load axis of the testing machine to ensure pure axial loading in the pullout test, while vertical displacement was the only degree of freedom in the cantilever bending test. The materials testing machine (Model 4302, Instron Inc., CantodMA, USA) was operated in displacement control using a cross-head speed of 0.082 m d s . The ultimate loads were determined from the load displacement curves recorded. In the case of the pullout tests the ultimate load was marked by a clear drop of the curve while a cutting-through effect with a stepwise increase of the load displacement curve was found under cantilever bending. Therefore the first maximum of the curve was taken as ultimate load of the screw in bending (Fig. 2). Determination of cortical thickness and cancellous bone density
The coordinates of the screws were determined by superimposing computer reconstructed X-rays from the CT data sets with the macroradiographs of the bone segments and steel cylinders taken prior to the testing. Using a custom made evaluation routine on basis of Matlab (Release 12, The Mathworks, Inc., NovilMI, USA) the alignment of the screws within the bone segments was determined and the density profile along each screw axis was computed using linear interpolation of the CT data (Fig. 3). Cortical thickness was determined by means of the gradient of the density (distance between maximum and minimum gradient) rather than the density profile itself. Due to the partial volume effect of CT scanners, a simple density
Fig. 1. Experimental set-up for the axial pullout (left) and the cantilever bending tests (right).
J. Seebeck et al. I Journal of Orthopedic Research 22 (2004) 1237-1242
*
thickness determined from the efficient thread length of 30 and 10 mm, respectively (efficient thread length = screw length - 5 mm head - 5 mm spacing - 6 mm tapping tip).
load [kN] 2.5
2.0
first maximum
Statistical model
cantilever bending
1.5
-
pull-out I
I
0.5
1 .o
I
I
1.5
2.0
b
displacement [mm] Fig. 2. Examples of load displacement curves for axial pullout and cantilever bending modes.
screw insertion depth cortical thickness
.3 1.5- a
1
cancellous insertion depth-
c
v)
a,
F
'
dmew
. linsert
The cortical as well as the cancellous insertion depth of the screw are therefore expected to have a linear influence on the ultimate holding strength of the screw. For the ultimate shear strength of cancellous bone a linear relationship to the bone density has been reported [S,l3]. In the case of cantilever bending, the bone is loaded by compression in its long axis. The compressive stresses can be estimated from the and the 2D protransverse forces applied at the screw head KranSYerSe jection of the screw shape calculated from the outer diameter of the screw d,, and their insertion depth linsenas 'Jcornpressive =
Ftransverse
dscrew ' linsert
The compressive stresses are therefore linearly correlated with the cortical thickness and the cancellous insertion depth. In a study of Ding et al. the ultimate compressive strength of cancellous bone from the proximal metaphysis of the tibia was also found to be linearly related to the apparent ash density, which is comparable to the bone mineral density values retrieved by CT values [6]. Therefore a linear correlation with the measured cancellous bone mineral density was expected and the statistical model formulated as follows: Fa,,,, = up . cort.th + b, . canc.dens + c, canc.insert
1.0-
firansverse = a, . cort.th + b, . canc.dens + cs . c a n c h e r t
2 0.5'E
E
2
&ai A
0.5
'0
In order to assess the importance of cortical thickness and bone density on the holding strength of internal fixator screws, a statistical model was used. This model rests on the following considerations. In axial loading, a screw will stress the bone interface in shear along the envelope cylindrical surface of the screw. The shear stresses can be calculated from the pullout force F&, the outer diameter of the screws d,,, and insertion depth linrertas Tshear =
1 .o
C
1239
0-
, 0
Fig. 3. CT-scan of tibia cross-section showing the density profile and bone parameters (cortical thickness, mean cancellous density, cancellous insertion depth) along the screw axis determined from the density profile (black) and the maximum (a) and minimum (b) of the density gradient profile (grey).
threshold could lead to either under or overestimation of cortical thickness [17,22]. Cancellous bone density was calculated as the average apparent bone density within the cancellous insertion depth. The cancellous insertion depth was calculated by subtracting the cortical
The experimental data was analysed by means of stepwise multiple linear regression including cortical thickness, cort.th; cancellous density, canc.dens and cancellous insertion depth, canc.insert. Coefficients up and a, described the influence of the cortical thickness on the screws respectively. Coefficients b, and 6, ultimate load F&l and Kransverse. showed the correlation to the cancellous density while coefficients c, and c, would reveal an influence of the cancellous insertion depth. Since neither pullout nor transverse screw forces can be expected if none of the parameters are existent, the regression model was defined with zero intercept. Density values for the cortical bone were not considered in the model, since differences in cortical density cannot be detected with sufficient accuracy by clinical CT scanners [23]. The importance of the cortical bone for screw fixation was therefore included in the cortical thickness alone. Statistical analysis was performed using SPSS for Windows (Version 10.0, SPSS Inc., Chicago/IL, USA). One way analysis of variance (ANOVA) was used to check for the influence of the screw head to bone distance on the failure load of the screw. A paired t-test was used to compare the magnitude of forces reached under an axial pullout and a cantilever bending between respective bone pairs. For all tests the confidence limit was set to 95%.
Results Values for cortical thickness and cancellous bone density are presented in Table 1. They showed a large variation between the individual bones and bone segments. Ultimate loads of a total of 144 screws were tested in a pullout loading (n = 72) and cantilever bending mode
J. Seebeck et al. I Journal of Orthopedic Research 22 (2004) 1237-1242
1240
Table I Cortical thickness and cancellous bone density Mean (SD)
Maximum
Minimum
n
2.83 (f1.15) 79.95 (f40.85)
6.12 169.88
0.74 5.08
111" 42b
Cortical thickness (mm) Apparent cancellous bone mineral density (mg/cm')
"Only including values of the meta and diaphyseal regions with an apparent cortical density >800 mg/cm3. Only including values of the metaphyseal bone segments.
Table 2 Mean ultimate screw loads per bone segment Bone segment Proximal
Distal
a Second
Axial pullout (kN)
Cantilever bending (kN)
I 2 3 4 5
0.263 k0.171 0.969 i0.382 0.797k 0.464 1.081 k0.438 1.486 i0.642
n = 13
8 9 10
1.328f 0.533 0.654 k 0.254 0.553 k 0.427
n = 12 n - 11
0.485f0.154 1.904f 0.710 1.857f0.716 2.415 2 1.005 2.6705 0.906
n=13 n=4 n = 1215" n = 12/9a n = 1219"
n=12 n=4 n=4
2.405 5 0.849 1.608 f 0.457 0.720 f 0.296
n = 1117"
n-4 n=
12
n=4 n=4
number indicates cases with screw breakage.
(n = 72) as shown in Table 2. In all cases of the axial pullout tests failure of the bone occurred. In the bending configuration not only the bone but also the screws failed in 30 tests. In these cases failure always occurred below the screw head. With a 5 mm head to bone distance screws failed before the bone, when forces reached about 2.4 kN. The ultimate load before breakage of the screws could be increased significantly when the distance between the screw head and the bone (free bending length) was reduced (Fig. 4). Multiple linear regression showed that the cortical thickness and the cancellous density had a significant influence on the ultimate holding strengths of the screws for both, pullout loading and cantilever bending modes. No significant influence was found for the cancellous insertion depths of the screws in the cancellous bone (Table 3). Comparison of the coefficients aP and a, for cortical thickness indicated that a screw engaged in cortices of the same thickness can take about twice as much transverse load when compared to axial pullout.
ultimate load [kN]
1
4.5.
-
1
5
3
distance between screw head and bone [mm]
Fig. 4. Reduction of failure load until screw breakage with increasing free bending length between screw head and bone in the cantilever bending mode (p < 0.001).
Table 3 Multiple linear regression models to predict ultimate axial and transverse forces
Axial force (kN)
Transverse force (kN)
Coefficient
n
f2
Cortical thickness (mm) Apparent cancellous bone mineral density (g/cm3) Cancellous insertion depth (mm)
up = 0.360* b, = 2.467
72
0.93
Cortical thickness (mm) Apparent cancellous bone mineral density (g/cm3) Cancellous insertion depth (mm)
a, = 0.681' b, = 5.888' nxa
3Ib
0.98
< 0.001. significant 3 excluded from the model. bOnly including screws with no breakage and a head to bone distance of 5 mm.
* Significant with p a Not
n.sa
J. Seebeck et al. I Journal of Orthopaedic Research 22 (2004) 1237-1242
A similar result was found when comparing the coefficients b, and b, for cancellous density. The bone could take about twice as much loading when submitted to transverse instead of axial loading. The stepwise multiple linear regression resulted in the following predictive formulas:
Faxial (kN) = 0.36 . cort.th (mm)
+ 2.47 . canc.dens (g/cm3) Ftransverse
(kN) = 0.68 . cort.th (mm)
+ 5.89. canc.dens (g/cm’) For 31 pairs of bone segments (Table 2) which allowed a direct paired comparison (screw head distance of 5 mm, no screw breakage), the paired t-test revealed a significant correlation between axial and transverse force (p < 0.001). The average force required for axial pullout was 0.51 kN (k0.35 kN) while in the cantilever bending mode 1.07 kN (k0.70 kN) were reached.
Discussion In this study we examined the influence of an axial pullout and cantilever bending mode acting at the screw head on the resulting holding power of the screw at different sites of the tibia. The results show that the loading of the screw in the cantilever bending mode leads to a significant increase of almost twice the amount in the holding power of the screw when compared to the axial loading situation. The anisotropic material properties of bone can be assumed to be an explanation for this observation [4,25]. By using the screw in a bending mode, like tested in this study, the bone is loaded along the direction of its highest compressive and tensile strength. An axial pullout however produces a loading of the bone in its weakest, transverse direction. That means, a screw which would fail in the axially loaded application of a plate osteosynthesis (e.g. in osteoporotic bone) might still provide enough holding power when it could be loaded in a cantilever bending mode, since it better uses the strength potential of the bone. For both load cases the holding power is dependent on the cortical thickness and the cancellous density. While for the cortical bone the insertion length defined by the cortical thickness was found to be the influencing parameter the anchorage strength seems to depend on the density of the cancellous born but not on the cancellous insertion depth of the screw. At first sight the latter is an unexpected result since others have shown that the pullout strength of a screw is also dependent on the insertion depth [3]. Those tests were, however, performed in homogeneous foam material, while in this
1241
study the screws were inserted into human bone specimens composed of both cortical and cancellous bone. The predictive formulas derived from this study are limited to the type and size of screws that have been used. However, other studies showed that bone density and the outer screw diameter have the biggest influence on screw holding strength, and only minor improvements can be reached with the modification of the thread design [1,5]. We therefore expect that the basic differences found in this study will not change for screws with a different thread design, since the effect of the thread shape will be minor compared to the influence of the loading mode, the cortical thickness and the cancellous density. The cases of screw breakage that occurred under cantilever bending show that the higher ultimate holding power achieved in this loading mode should also be reflected in the design of the screws for use with internal fixators. Based on the statistical evaluation we can predict that a bone site with a cortical thickness of more than 5 mm will only be loaded up to 70% of its anchorage capacity when the screw fails. However, loads of such a critical magnitude might not be reached in the clinical application, since no screw breakage has been reported so far [19]. It could be shown that a reduction of the free bending length of the screws between the head in the fixator and the bone surface will significantly reduce the risk of screw breakage. This indicates, that bending rather than shear, both induced in a cantilever bending mode, is the critical factor for screw failure. Whether the free bending length might also influence the load bearing at the bone-screw interface could not be answered in this study. In cases with comparable bone morphology the screws failed before the bone and no comparison between the tests with different free bending lengths was possible. One could however presume that a larger bending of the screw due to a larger free bending length also leads to a higher loading of the bone interface. In this context dynamic loosening of the screw has to be considered, although it was not addressed in this study. An increased loading of the bone might lead to a progressive increase of the diameter of the bone interface under cyclic loading, resulting in a loosening of the screw and failure of the anchorage. Overall it can be concluded that a good fit of an internal fixator to the shape of the bone surface should be aspired in order to reduce the free bending lengths of the screws and decrease the risk of screw breakage or loosening. Based on the findings of this study the desirable osteosynthesis would be an internal fixator designed and applied to the bone in such a way that the screws are loaded in a pure bending mode and the long bone in its longitudinal direction. Obviously this claim cannot generally be achieved due to the complexity of anatomy and physiological muscle and joint loading. But it might
1242
J. Seebeck et al. I Journal of’Orthopaedic Research 22 (2004) 1237-1242
be useful to further analyse the loads acting on the screws used with an internal fixator at different anatomical locations under different physiological loads and different fracture patterns. Besides their obvious biological advantages, internal h a t o r s submit their screws to a cantilever bending mode which better uses the direction-dependent mechanical properties of the bone. Therefore these implants seems to have a great mechanical potential for application in long bones, especially in osteoporotic patients, when the mechanical strength of the bone is low.
Acknowledgements The authors gratefully acknowledge Mathys Medical Ltd., Bettlach, Switzerland for providing the screws for the experiments.
References [I] Asnis SE, Ernberg JJ, Bostrom MP, Wright TM, Harrington RM, Tencer A, et al. Cancellous bone screw thread design and holding power. J Orthop Trauma 1996;10:462-9. [2] Brown GA, McCarthy T, Bourgeault CA, Callahan DJ. Mechanical performance of standard and cannulated 4.0 mm cancellous bone screws. J Orthop Res 2000;18:307-12. [3] Chapman JR, Harrington RM, Lee KM, Anderson PA, Tencer AF, Kowalski D. Factors affecting the pullout strength of cancellous bone screws. J Biomech Eng 1996;118:391-8. [4] Ciarelli MJ, Goldstein SA, Kuhn JL, Cody DD, Brown MB. Evaluation of orthogonal mechanical properties and density of human trabecular bone from the major metaphyseal regions with materials testing and computed tomography. J Orthop Res 1991;9:67482. [5] DeCoster TA, Heetderks DB, Downey DJ, Ferries JS, Jones W. Optimizing bone screw pullout force. J Orthop Trauma 1990; 4:169-74. [6] Ding M, Dalstra M, Danielsen CC, Kabel J, Hvid I, Linde F. Age variations in the properties of human tibia1 trabecular bone. J Bone Joint Surg Br 1997;79:995-1002. [7] Faran KJ, lchioka N, Trzeciak MA, Han S, Medige J, Moy OJ. Effect of bone quality on the forces generated by compression screws. J Biomech 1999;32:8614. [8] Ford CM, Keaveny TM. The dependence of shear failure properties of trabecular bone on apparent density and trabecular orientation. J Biomech 1996;29:1309-17. [9] Frigg R. Locking compression plate (LCP). An osteosynthesis plate based on the dynamic compression plate and the point contact fixator (PC-Fix). Injury 2001;32(Suppl 2):63-6.
[lo] Harnroongroj T, Techataweewan A. Determination of the role of the cancellous bone in generation of screw holding power at metaphysis. Clin Biomech (Bristol, Avon) 1999;14:3644. [Ill Horsman A, Leung WK, Bentley HB, McLachlan MS. Effect of rotation on radiographic dimensions of the humerus and femur. Br J Radio1 1977;50:23-8. [12] Messmer P, Long G, Suhm N, Regazzoni P, Jacob AL. Volumetric model determination of the tibia based on 2D radiographs using a 2D13D database. Comput Aided Surg 2001;6:183-94. [13] Mitton D, Cendre E, Roux JP, Arlot ME, Peix G, Rumelhart C, et al. Mechanical properties of ewe vertebral cancellous bone compared with histomorphometry and high-resolution computed tomography parameters. Bone 1998;22:651-8. Pearson J, Ruegsegger P, Dequeker J, Henley M, Bright J, Reeve J, et al. European semi-anthropomorphic phantom for the crosscalibration of peripheral bone densitometers: assessment of precision accuracy and stability. Bone Miner 1994;27:109-20. Perren SM. Evolution and rationale of locked internal fixator technology. Introductory remarks. Injury 2001;32(Suppl 2):S-9. Perren SM, Buchanan JS. Basic concepts relevant to the design and development of the Point Contact Fixator (PC-Fix). Injury 1995;26(S~ppl2):S-BI-B4. Ruegsegger P, Munch B, Felder M. Early detection of osteoarthritis by 3D computed tomography. Techno1 Health Care 1993;1:53M6. Schatzker J, Sanderson R, Murnaghan JP. The holding power of orthopedic screws in vivo. Clin Orthop 1975:115-26. Schuetz M, Mueller M, Krettek C, Hoentzsch D, Regazzoni P, Ganz R, et al. Minimally invasive fracture stabilization of distal femoral fractures with the LISS: A prospective multicenter study. Results of a clinical study with special emphasis on difficult cases. Injury 2001;32:S-C48-C54. Seebeck J, Lill C, Morlock MM, Schneider E. Implant fixation in osteoporotic bone using monocortical screws. Trauma Berufskrankh 1999;1:4114. Seebeck J, Morlock MM, Schneider E. Optimal implant anchorage in osteoporotic bone using unicortical screws. In: Transactions of the 46th Annual Meeting, Orthopaedic Research Society. 2000, p. 879. Seitz P, Ruegsegger P. Fast contour detection algorithm for high precision quantitative CT. IEEE Trans Med Imag 1983;M1-2:1 3 6 41. Snyder SM, Schneider E. Estimation of mechanical properties of cortical bone by computed tomography. J Orthop Res 1991;9:42231. Stromsoe K, Kok WL, Hoiseth A, Alho A. Holding power of the 4.5 mm AO/ASIF cortex screw in cortical bone in relation to bone mineral. Injury 1993;246569. Wirtz DC, Schiffers N, Pandorf T, Radermacher K, Weichert D, Forst R. Critical evaluation of known bone material properties to realize anisotropic FE-simulation of the proximal femur. J Biomech 2000;33:1325-30. Zand MS, Goldstein SA, Matthews LS. Fatigue failure of cortical bone screws. J Biomech 1983;16:305-11.
Journal of Orthopaedic Research
Journal of Orthopaedic Research 22 (2004) 1237-1242
www.elsevier.com/locate/orthres
Effect of cortical thickness and cancellous bone density on the holding strength of internal fixator screws J. Seebeck a, J. Goldhahn a, H. Stadele b, P. Messmer ‘, M.M. Morlock ‘, E. Schneider a
A 0 Research Institute. Davos, Clavadelerstrasse, CH-7270 Davos Platz, Switzerland CA N C A S Research Group, University Hospital, Basel, Switzerland Biomechanics Section, Technical University Hamburg-Harburg, Germany
Abstract Internal fixators are a new class of implants designed to preserve the periosteal blood supply of the bone. In contrast to conventional plate fixation in which the screws have spherical heads and are loaded mainly by axial pullout forces, screws in internal fixators are “locked” within the plate and therefore subjected to axial as well as bending loads. In this study the ultimate loads of screws of a commercially available internal fixator system were tested in a pullout (n = 72) and cantilever bending mode (n = 72) in metaphyseal and diaphyseal regions of four pairs of human tibiae with different bone qualities. Cortical thickness and cancellous bone density were determined at the screw insertion sites. Stepwise multiple linear regression revealed that cortical thickness and cancellous density can explain 93% and 98% of the variance of the ultimate load of the screws in an axial pullout and cantilever bending mode. Screws in internal fixators are better suited to transmit shear forces and thereby make better use of the strength potential of bone than screws used in conventional plate fixation: this is especially advantageous when bone strength is reduced, e.g. due to osteoporosis. 0 2004 Published by Elsevier Ltd. on behalf of Orthopaedic Research Society. Keywords: Screw fixation; Internal fixation; Bone morphology; Fracture treatment
Introduction Screws are an essential element in fracture fixation in the metaphyseal and diaphyseal regions of bone. The maximum load that can be transferred between a screw and the bone is in many cases the limiting factor for the success of the osteosynthesis. In the past, the holding strengths of different screws have been investigated by means of axial pullout tests in cortical and cancellous bone specimens as well as synthetic foams [l-3,5,7, 10,18,24]. The testing of screws in the axial direction was derived from their principal mode of operation in conventional plate fixation. The screws are axially preloaded to generate compression between bone and plate. This compression should result in sufficient friction to transfer loads that act orthogonal to the screw axis. In plate fixation the screw is not supposed to take *Corresponding author. Tel.: +41-81-4142441; fax: +41-814142288. E-mail address: [email protected] (E. Schneider).
significant bending loads to avoid shear failure [26]. This agrees with the common understanding on the usage of screws in mechanical engineering. Some years ago the concept of internal fixators with unicortical screws for treatment of long bone fractures was introduced [15,16]. The initial motivation to develop these types of implants was based on the biological advantage of saving the periosteal blood supply of the bone by avoiding compression between plate and bone. In these internal fixators mechanical stability is maintained by rigidly locking the heads of the unicortical screws within the plate. In this configuration the screws are no longer used in the classical way but act mechanically like the pins of an external fixator. As a consequence the screws are now sustaining a pronounced cantilever bending in addition to axial loads 1161. In a previous study using bovine cortical bone specimens we showed that a screw submitted to bending can transmit two times the loads tolerated in axial pullout before failure of the bone [20,21]. However, no information is available regarding the amount of load that
0736-0266/$ - see front matter 0 2004 Published by Elsevier Ltd. on behalf of Orthopaedic Research Society. doi:10.1016/j.orthres.2004.04.001
1238
J. Seebeck et al. I Journal of Orthopaedic Research 22 (2004) 1237-1242
can be transferred in the metaphyseal and diaphyseal regions of human bone when the screw is subjected to bending vs. axial loads. In this context, the cortical thickness and the cancellous bone density are of importance since insufficient screw anchorage is a serious problem in bone that has become mechanically weak as e.g. in osteoporosis. The purpose of this study was to determine the holding strength of screws used in an internal fixator system in both, an axial pullout and a cantilever bending mode at different sites of the human tibia and to thereby evaluate the influence of cortical thickness and bone density on the holding strength of the internal fixator screws.
Materials and methods Specimen preparation
The cortical index (CI), i.e. the ratio between the cortical thickness and the bone radius in the central cross-section of the diaphysis was determined by means of an antero-posterior X-ray in a pool of 10 pairs of fresh frozen human tibiae [ll]. Two pairs (61 years-old female, 70 years-old male) with the highest CI (0.41,0.48) and two pairs (82 yearsold female, 72 years-old male) with the lowest CI (0.19, 0.23) were chosen. These eight bones were scanned in a clinical CT-Scanner (Somatom 4 volume zoom, Siemens, Erlangen, Germany) using a slice thickness and slice distance of 1 mm each, and an in-plane resolution of 0.234 mm (scan diameter 120 mm, picture size 512x 512 pixels, 4 x 1 mm collimation, 120 kV, 100 mA) according to a previously described setup [12]. A phantom (European Forearm Phantom) [I41 was included in the scan of each bone to later relate the absorption values of the cross-sectional CT images with the apparent bone mineral density. The tibiae were divided into 10 sections of equal lengths. Five proximal and the three most distal bone segments were prepared by transversely cutting each tibia by means of a diamond band saw (Exakt, Norderstedt, Germany). Screw insertion
Commercially available titanium screws (Synthes, LCP-Locking Head Screws 05.0 mm, lengths 46 and 26 mm, Mathys Medical Ltd., Bettlach, Switzerland) were used for both pullout and cantilever bending tests. The screws, designed to be used in both cortical and cancellous bone are self-tapping and have a core diameter of 4.4 mm, an outer diameter of 5 mm and a pitch of 1 mm. The heads of the screws are formed as a conical thread which locks the screw within in the corresponding fixator system [9]. For the pullout and cantilever bending tests a mounting block with the same conical thread replaced the fixator. The number of screws that could be placed into each bone segment was limited by the sizes of the segments. Making sure that each screw did not interfere with the others, three screws were placed in the first proximal metaphyseal segment of all bones except for one pair of bones that was of bigger size and allowed a placement of four screws. In the second proximal segment as well as in the last two distal segments just one screw could be tested in all bones. The remaining diaphyseal bone segments allowed a placement of three screws per segment in accordance with the triangular cross-sectional shape of the tibia1 diaphysis. The screws in the first two proximal segments as well as the last distal segment had a total length of 46 mm while in the other segments screws with a length of 26 mm were used to achieve a unicortical insertion as suggested for the use with internal fixators. In accordance with the surgical technique recommended for this type of screw, 4.3 mm pilot holes were drilled with an insertion point at 50% height of the bone segments and with a depth corresponding to the screw insertion length. A linearly guided machine drill was used
such that the drill axes were adjusted parallel to the segment cutting planes to ensure co-linearity of the screw axes with the distraction force under axial pullout and pure transverse loading in the cantilever bending mode. The self-tapping screws were inserted by hand holding the mounting block in place until the conical head of the screw was locked into it. The size of the mounting block only allowed a sequential insertion and testing of each single screw. In order to record the position and direction of the screw axes 4.0 mm stainless steel cylinders were placed into the predrilled holes before screw insertion and macroradiographs of the bone segments with the inserted cylinders were taken in the proximo-distal direction (cabinet X-ray system, model 43855A, Faxitron X-ray Corporation, IL, USA). Pullout and cantilever bending tests
Screws were axially pulled out from one bone of each pair while in the counterpart the screws were loaded in a cantilever bending mode pulling in the direction of the long axis of the bone (Fig. 1). In the cantilever bending mode, the free bending length of the screw between the locked head and the bone determines the bending moment induced into the screw and has an influence on the load bearing at the bonescrew interface. Therefore the free bending length was kept constant by placing a spacer of 5 mm thickness between the mounting block and the bone surface during screw insertion. In addition, cantilever bending tests of screws with free bending lengths of 1 and 3 mm were performed for 20 specimens to determine the effect of the free bending length. In both loading modes, the bone segments were rigidly connected to the base plate. The midpoint of the screw heads were aligned with the load axis of the testing machine to ensure pure axial loading in the pullout test, while vertical displacement was the only degree of freedom in the cantilever bending test. The materials testing machine (Model 4302, Instron Inc., CantodMA, USA) was operated in displacement control using a cross-head speed of 0.082 m d s . The ultimate loads were determined from the load displacement curves recorded. In the case of the pullout tests the ultimate load was marked by a clear drop of the curve while a cutting-through effect with a stepwise increase of the load displacement curve was found under cantilever bending. Therefore the first maximum of the curve was taken as ultimate load of the screw in bending (Fig. 2). Determination of cortical thickness and cancellous bone density
The coordinates of the screws were determined by superimposing computer reconstructed X-rays from the CT data sets with the macroradiographs of the bone segments and steel cylinders taken prior to the testing. Using a custom made evaluation routine on basis of Matlab (Release 12, The Mathworks, Inc., NovilMI, USA) the alignment of the screws within the bone segments was determined and the density profile along each screw axis was computed using linear interpolation of the CT data (Fig. 3). Cortical thickness was determined by means of the gradient of the density (distance between maximum and minimum gradient) rather than the density profile itself. Due to the partial volume effect of CT scanners, a simple density
Fig. 1. Experimental set-up for the axial pullout (left) and the cantilever bending tests (right).
J. Seebeck et al. I Journal of Orthopedic Research 22 (2004) 1237-1242
*
thickness determined from the efficient thread length of 30 and 10 mm, respectively (efficient thread length = screw length - 5 mm head - 5 mm spacing - 6 mm tapping tip).
load [kN] 2.5
2.0
first maximum
Statistical model
cantilever bending
1.5
-
pull-out I
I
0.5
1 .o
I
I
1.5
2.0
b
displacement [mm] Fig. 2. Examples of load displacement curves for axial pullout and cantilever bending modes.
screw insertion depth cortical thickness
.3 1.5- a
1
cancellous insertion depth-
c
v)
a,
F
'
dmew
. linsert
The cortical as well as the cancellous insertion depth of the screw are therefore expected to have a linear influence on the ultimate holding strength of the screw. For the ultimate shear strength of cancellous bone a linear relationship to the bone density has been reported [S,l3]. In the case of cantilever bending, the bone is loaded by compression in its long axis. The compressive stresses can be estimated from the and the 2D protransverse forces applied at the screw head KranSYerSe jection of the screw shape calculated from the outer diameter of the screw d,, and their insertion depth linsenas 'Jcornpressive =
Ftransverse
dscrew ' linsert
The compressive stresses are therefore linearly correlated with the cortical thickness and the cancellous insertion depth. In a study of Ding et al. the ultimate compressive strength of cancellous bone from the proximal metaphysis of the tibia was also found to be linearly related to the apparent ash density, which is comparable to the bone mineral density values retrieved by CT values [6]. Therefore a linear correlation with the measured cancellous bone mineral density was expected and the statistical model formulated as follows: Fa,,,, = up . cort.th + b, . canc.dens + c, canc.insert
1.0-
firansverse = a, . cort.th + b, . canc.dens + cs . c a n c h e r t
2 0.5'E
E
2
&ai A
0.5
'0
In order to assess the importance of cortical thickness and bone density on the holding strength of internal fixator screws, a statistical model was used. This model rests on the following considerations. In axial loading, a screw will stress the bone interface in shear along the envelope cylindrical surface of the screw. The shear stresses can be calculated from the pullout force F&, the outer diameter of the screws d,,, and insertion depth linrertas Tshear =
1 .o
C
1239
0-
, 0
Fig. 3. CT-scan of tibia cross-section showing the density profile and bone parameters (cortical thickness, mean cancellous density, cancellous insertion depth) along the screw axis determined from the density profile (black) and the maximum (a) and minimum (b) of the density gradient profile (grey).
threshold could lead to either under or overestimation of cortical thickness [17,22]. Cancellous bone density was calculated as the average apparent bone density within the cancellous insertion depth. The cancellous insertion depth was calculated by subtracting the cortical
The experimental data was analysed by means of stepwise multiple linear regression including cortical thickness, cort.th; cancellous density, canc.dens and cancellous insertion depth, canc.insert. Coefficients up and a, described the influence of the cortical thickness on the screws respectively. Coefficients b, and 6, ultimate load F&l and Kransverse. showed the correlation to the cancellous density while coefficients c, and c, would reveal an influence of the cancellous insertion depth. Since neither pullout nor transverse screw forces can be expected if none of the parameters are existent, the regression model was defined with zero intercept. Density values for the cortical bone were not considered in the model, since differences in cortical density cannot be detected with sufficient accuracy by clinical CT scanners [23]. The importance of the cortical bone for screw fixation was therefore included in the cortical thickness alone. Statistical analysis was performed using SPSS for Windows (Version 10.0, SPSS Inc., Chicago/IL, USA). One way analysis of variance (ANOVA) was used to check for the influence of the screw head to bone distance on the failure load of the screw. A paired t-test was used to compare the magnitude of forces reached under an axial pullout and a cantilever bending between respective bone pairs. For all tests the confidence limit was set to 95%.
Results Values for cortical thickness and cancellous bone density are presented in Table 1. They showed a large variation between the individual bones and bone segments. Ultimate loads of a total of 144 screws were tested in a pullout loading (n = 72) and cantilever bending mode
J. Seebeck et al. I Journal of Orthopedic Research 22 (2004) 1237-1242
1240
Table I Cortical thickness and cancellous bone density Mean (SD)
Maximum
Minimum
n
2.83 (f1.15) 79.95 (f40.85)
6.12 169.88
0.74 5.08
111" 42b
Cortical thickness (mm) Apparent cancellous bone mineral density (mg/cm')
"Only including values of the meta and diaphyseal regions with an apparent cortical density >800 mg/cm3. Only including values of the metaphyseal bone segments.
Table 2 Mean ultimate screw loads per bone segment Bone segment Proximal
Distal
a Second
Axial pullout (kN)
Cantilever bending (kN)
I 2 3 4 5
0.263 k0.171 0.969 i0.382 0.797k 0.464 1.081 k0.438 1.486 i0.642
n = 13
8 9 10
1.328f 0.533 0.654 k 0.254 0.553 k 0.427
n = 12 n - 11
0.485f0.154 1.904f 0.710 1.857f0.716 2.415 2 1.005 2.6705 0.906
n=13 n=4 n = 1215" n = 12/9a n = 1219"
n=12 n=4 n=4
2.405 5 0.849 1.608 f 0.457 0.720 f 0.296
n = 1117"
n-4 n=
12
n=4 n=4
number indicates cases with screw breakage.
(n = 72) as shown in Table 2. In all cases of the axial pullout tests failure of the bone occurred. In the bending configuration not only the bone but also the screws failed in 30 tests. In these cases failure always occurred below the screw head. With a 5 mm head to bone distance screws failed before the bone, when forces reached about 2.4 kN. The ultimate load before breakage of the screws could be increased significantly when the distance between the screw head and the bone (free bending length) was reduced (Fig. 4). Multiple linear regression showed that the cortical thickness and the cancellous density had a significant influence on the ultimate holding strengths of the screws for both, pullout loading and cantilever bending modes. No significant influence was found for the cancellous insertion depths of the screws in the cancellous bone (Table 3). Comparison of the coefficients aP and a, for cortical thickness indicated that a screw engaged in cortices of the same thickness can take about twice as much transverse load when compared to axial pullout.
ultimate load [kN]
1
4.5.
-
1
5
3
distance between screw head and bone [mm]
Fig. 4. Reduction of failure load until screw breakage with increasing free bending length between screw head and bone in the cantilever bending mode (p < 0.001).
Table 3 Multiple linear regression models to predict ultimate axial and transverse forces
Axial force (kN)
Transverse force (kN)
Coefficient
n
f2
Cortical thickness (mm) Apparent cancellous bone mineral density (g/cm3) Cancellous insertion depth (mm)
up = 0.360* b, = 2.467
72
0.93
Cortical thickness (mm) Apparent cancellous bone mineral density (g/cm3) Cancellous insertion depth (mm)
a, = 0.681' b, = 5.888' nxa
3Ib
0.98
< 0.001. significant 3 excluded from the model. bOnly including screws with no breakage and a head to bone distance of 5 mm.
* Significant with p a Not
n.sa
J. Seebeck et al. I Journal of Orthopaedic Research 22 (2004) 1237-1242
A similar result was found when comparing the coefficients b, and b, for cancellous density. The bone could take about twice as much loading when submitted to transverse instead of axial loading. The stepwise multiple linear regression resulted in the following predictive formulas:
Faxial (kN) = 0.36 . cort.th (mm)
+ 2.47 . canc.dens (g/cm3) Ftransverse
(kN) = 0.68 . cort.th (mm)
+ 5.89. canc.dens (g/cm’) For 31 pairs of bone segments (Table 2) which allowed a direct paired comparison (screw head distance of 5 mm, no screw breakage), the paired t-test revealed a significant correlation between axial and transverse force (p < 0.001). The average force required for axial pullout was 0.51 kN (k0.35 kN) while in the cantilever bending mode 1.07 kN (k0.70 kN) were reached.
Discussion In this study we examined the influence of an axial pullout and cantilever bending mode acting at the screw head on the resulting holding power of the screw at different sites of the tibia. The results show that the loading of the screw in the cantilever bending mode leads to a significant increase of almost twice the amount in the holding power of the screw when compared to the axial loading situation. The anisotropic material properties of bone can be assumed to be an explanation for this observation [4,25]. By using the screw in a bending mode, like tested in this study, the bone is loaded along the direction of its highest compressive and tensile strength. An axial pullout however produces a loading of the bone in its weakest, transverse direction. That means, a screw which would fail in the axially loaded application of a plate osteosynthesis (e.g. in osteoporotic bone) might still provide enough holding power when it could be loaded in a cantilever bending mode, since it better uses the strength potential of the bone. For both load cases the holding power is dependent on the cortical thickness and the cancellous density. While for the cortical bone the insertion length defined by the cortical thickness was found to be the influencing parameter the anchorage strength seems to depend on the density of the cancellous born but not on the cancellous insertion depth of the screw. At first sight the latter is an unexpected result since others have shown that the pullout strength of a screw is also dependent on the insertion depth [3]. Those tests were, however, performed in homogeneous foam material, while in this
1241
study the screws were inserted into human bone specimens composed of both cortical and cancellous bone. The predictive formulas derived from this study are limited to the type and size of screws that have been used. However, other studies showed that bone density and the outer screw diameter have the biggest influence on screw holding strength, and only minor improvements can be reached with the modification of the thread design [1,5]. We therefore expect that the basic differences found in this study will not change for screws with a different thread design, since the effect of the thread shape will be minor compared to the influence of the loading mode, the cortical thickness and the cancellous density. The cases of screw breakage that occurred under cantilever bending show that the higher ultimate holding power achieved in this loading mode should also be reflected in the design of the screws for use with internal fixators. Based on the statistical evaluation we can predict that a bone site with a cortical thickness of more than 5 mm will only be loaded up to 70% of its anchorage capacity when the screw fails. However, loads of such a critical magnitude might not be reached in the clinical application, since no screw breakage has been reported so far [19]. It could be shown that a reduction of the free bending length of the screws between the head in the fixator and the bone surface will significantly reduce the risk of screw breakage. This indicates, that bending rather than shear, both induced in a cantilever bending mode, is the critical factor for screw failure. Whether the free bending length might also influence the load bearing at the bone-screw interface could not be answered in this study. In cases with comparable bone morphology the screws failed before the bone and no comparison between the tests with different free bending lengths was possible. One could however presume that a larger bending of the screw due to a larger free bending length also leads to a higher loading of the bone interface. In this context dynamic loosening of the screw has to be considered, although it was not addressed in this study. An increased loading of the bone might lead to a progressive increase of the diameter of the bone interface under cyclic loading, resulting in a loosening of the screw and failure of the anchorage. Overall it can be concluded that a good fit of an internal fixator to the shape of the bone surface should be aspired in order to reduce the free bending lengths of the screws and decrease the risk of screw breakage or loosening. Based on the findings of this study the desirable osteosynthesis would be an internal fixator designed and applied to the bone in such a way that the screws are loaded in a pure bending mode and the long bone in its longitudinal direction. Obviously this claim cannot generally be achieved due to the complexity of anatomy and physiological muscle and joint loading. But it might
1242
J. Seebeck et al. I Journal of’Orthopaedic Research 22 (2004) 1237-1242
be useful to further analyse the loads acting on the screws used with an internal fixator at different anatomical locations under different physiological loads and different fracture patterns. Besides their obvious biological advantages, internal h a t o r s submit their screws to a cantilever bending mode which better uses the direction-dependent mechanical properties of the bone. Therefore these implants seems to have a great mechanical potential for application in long bones, especially in osteoporotic patients, when the mechanical strength of the bone is low.
Acknowledgements The authors gratefully acknowledge Mathys Medical Ltd., Bettlach, Switzerland for providing the screws for the experiments.
References [I] Asnis SE, Ernberg JJ, Bostrom MP, Wright TM, Harrington RM, Tencer A, et al. Cancellous bone screw thread design and holding power. J Orthop Trauma 1996;10:462-9. [2] Brown GA, McCarthy T, Bourgeault CA, Callahan DJ. Mechanical performance of standard and cannulated 4.0 mm cancellous bone screws. J Orthop Res 2000;18:307-12. [3] Chapman JR, Harrington RM, Lee KM, Anderson PA, Tencer AF, Kowalski D. Factors affecting the pullout strength of cancellous bone screws. J Biomech Eng 1996;118:391-8. [4] Ciarelli MJ, Goldstein SA, Kuhn JL, Cody DD, Brown MB. Evaluation of orthogonal mechanical properties and density of human trabecular bone from the major metaphyseal regions with materials testing and computed tomography. J Orthop Res 1991;9:67482. [5] DeCoster TA, Heetderks DB, Downey DJ, Ferries JS, Jones W. Optimizing bone screw pullout force. J Orthop Trauma 1990; 4:169-74. [6] Ding M, Dalstra M, Danielsen CC, Kabel J, Hvid I, Linde F. Age variations in the properties of human tibia1 trabecular bone. J Bone Joint Surg Br 1997;79:995-1002. [7] Faran KJ, lchioka N, Trzeciak MA, Han S, Medige J, Moy OJ. Effect of bone quality on the forces generated by compression screws. J Biomech 1999;32:8614. [8] Ford CM, Keaveny TM. The dependence of shear failure properties of trabecular bone on apparent density and trabecular orientation. J Biomech 1996;29:1309-17. [9] Frigg R. Locking compression plate (LCP). An osteosynthesis plate based on the dynamic compression plate and the point contact fixator (PC-Fix). Injury 2001;32(Suppl 2):63-6.
[lo] Harnroongroj T, Techataweewan A. Determination of the role of the cancellous bone in generation of screw holding power at metaphysis. Clin Biomech (Bristol, Avon) 1999;14:3644. [Ill Horsman A, Leung WK, Bentley HB, McLachlan MS. Effect of rotation on radiographic dimensions of the humerus and femur. Br J Radio1 1977;50:23-8. [12] Messmer P, Long G, Suhm N, Regazzoni P, Jacob AL. Volumetric model determination of the tibia based on 2D radiographs using a 2D13D database. Comput Aided Surg 2001;6:183-94. [13] Mitton D, Cendre E, Roux JP, Arlot ME, Peix G, Rumelhart C, et al. Mechanical properties of ewe vertebral cancellous bone compared with histomorphometry and high-resolution computed tomography parameters. Bone 1998;22:651-8. Pearson J, Ruegsegger P, Dequeker J, Henley M, Bright J, Reeve J, et al. European semi-anthropomorphic phantom for the crosscalibration of peripheral bone densitometers: assessment of precision accuracy and stability. Bone Miner 1994;27:109-20. Perren SM. Evolution and rationale of locked internal fixator technology. Introductory remarks. Injury 2001;32(Suppl 2):S-9. Perren SM, Buchanan JS. Basic concepts relevant to the design and development of the Point Contact Fixator (PC-Fix). Injury 1995;26(S~ppl2):S-BI-B4. Ruegsegger P, Munch B, Felder M. Early detection of osteoarthritis by 3D computed tomography. Techno1 Health Care 1993;1:53M6. Schatzker J, Sanderson R, Murnaghan JP. The holding power of orthopedic screws in vivo. Clin Orthop 1975:115-26. Schuetz M, Mueller M, Krettek C, Hoentzsch D, Regazzoni P, Ganz R, et al. Minimally invasive fracture stabilization of distal femoral fractures with the LISS: A prospective multicenter study. Results of a clinical study with special emphasis on difficult cases. Injury 2001;32:S-C48-C54. Seebeck J, Lill C, Morlock MM, Schneider E. Implant fixation in osteoporotic bone using monocortical screws. Trauma Berufskrankh 1999;1:4114. Seebeck J, Morlock MM, Schneider E. Optimal implant anchorage in osteoporotic bone using unicortical screws. In: Transactions of the 46th Annual Meeting, Orthopaedic Research Society. 2000, p. 879. Seitz P, Ruegsegger P. Fast contour detection algorithm for high precision quantitative CT. IEEE Trans Med Imag 1983;M1-2:1 3 6 41. Snyder SM, Schneider E. Estimation of mechanical properties of cortical bone by computed tomography. J Orthop Res 1991;9:42231. Stromsoe K, Kok WL, Hoiseth A, Alho A. Holding power of the 4.5 mm AO/ASIF cortex screw in cortical bone in relation to bone mineral. Injury 1993;246569. Wirtz DC, Schiffers N, Pandorf T, Radermacher K, Weichert D, Forst R. Critical evaluation of known bone material properties to realize anisotropic FE-simulation of the proximal femur. J Biomech 2000;33:1325-30. Zand MS, Goldstein SA, Matthews LS. Fatigue failure of cortical bone screws. J Biomech 1983;16:305-11.