Iliotibial Band Tenodesis: A New Strategy for Attachment*

Louis F.

Draganich,† PhD, Yeou-Fang Hsieh, MS,

and Bruce Reider, MD

From The

University of Chicago Medical Center, Department of Surgery, Section of Orthopaedics and Rehabilitation Medicine, Chicago, Illinois
The extraarticular reconstruction has been proposed either as a substitute 2,3,5,20 or as an adjunct6,9,12,15,22,25 to the intrarticular reconstruction of the ACL-deficient knee. In recent years, extraarticular reconstruction has largely been abandoned for several reasons. Clinical studies 3, 11 suggest that the extraarticular reconstruction stretches with time. In vitro research in our laboratory, and in other laboratories,19 has demonstrated the extraarticular reconstruction alone to be incapable of returning normal anterior stability to the ACL-deficient knee. Also, the success of the intraarticular reconstruction through arthroscopic surgery has made clinicians reluctant to add the open surgery necessary to perform the extraarticular reconstruction. However, failures of intraarticular grafts still occur. Recent clinical studies have found more than 20% of patients with an intraarticular graft to have 3 to 6 mm or more of anterior drawer motion in the reconstructed knee compared with the normal contralateral knee.21,23 In addition, a recent clinical study found that of 82 consecutive patients classified as having ACL-deficient knees, iliotibial tract injuries of the knee occurred in 93% of those and correlated highly with variations in grades of anterior translation and pivot shift.2 Only 11% of the iliotibial tract lesions occurred to the superficial layers of the iliotibial tract, leaving it available for tenodesis in most instances. Furthermore, a recent cadaveric study found increased anterior translation in both flexion and extension of the knee and also found increased internal rotation at 90 of flexion with combined injuries to the ACL and anterolateral structures of the knee .2 These studies suggest that knees with large pathologic increases in anterolateral laxity have injured anterolateral structures in addition to the ACL. This implies a role for adjunctive lateral extraarticular reconstruction in a severely unstable knee if an effective technique could be

ABSTRACT
We

investigated the changes in distance between Gerdys tubercle on the tibia and points on the posterior

two thirds of the lateral surface of the lateral femoral

femoral shaft in 15 cadaveric knees. A three-dimensional digitizer was used to quantify motion of the knee during flexion ranging from full extension to 120° of flexion. Four load states were applied: internal, external, and neutral rotation, and quadriceps muscles loads based on one third of values in the literature for maximal isometric quadriceps muscles moments. The femoral location most isometric to Gerdys tubercle was found to be strongly influenced by the load state. A 1.0 cm wide iliotibial band tenodesis was modelled by five straight lines arising from Gerdys tubercle and attaching to a simulated washer at the junction of the lateral femoral condyle and shaft. Using this model and the motion data obtained from the cadavers, we investigated the effects of quadriceps muscles loading and external rotation of the knee on changes in the distances between these tibial and femoral attachments for each of the five lines. A 180° twist modelled into the tenodesis significantly reduced the range of changes in distance (difference between the largest and smallest changes in distance among the lines for a given angle of flexion) for both of these load states. Therefore, a 180° twist in the tenodesis can enhance isometry among the fibers of the tenodesis. This implies that a 180° twist can enhance load sharing among the fibers of the tenodesis and, therefore, enhance the overall strength of the tenodesis.

condyle and adjacent lateral

performed.
One of the most popular sites for attachment of the iliotibial band tenodesis has been Krackows point F9 at the junction of the lateral femoral condyle and lateral femoral shaft.&dquo; The tenodesis has generally been pulled tight during surgery, thereby externally rotating the tibia. 21 In two of the most extensive cadaveric studies on potential attachment sites for the iliotibial band tenodesis, very low, nonphysiologic quadriceps muscles loads were applied.ls>24 In
186

Presented in part at the 40th annual meetmg of the Orthopaedic Research New Orleans, Louisiana, February 1994. t Address correspondence and repnnt requests to Louis F. Draganich, PhD, The Umversity of Chicago Medical Center, Department of Surgery, Section of Orthopaedics and Rehabilitation Medicine, 5841 South Maryland Avenue, Chicago, IL 60637. No author or related mstitution has received any financial benefit from research m this study. See &dquo;Acknowledgments&dquo; for funding information.

Society,

187

of those studies, 16 additional load states were not tested, and in the other study,24 only three independent samples were used to determine the effect of load on isometry. In previous studies on isometry, the iliotibial band tenodesis was modelled as a line with single point-to-point attachments.16,24 Our preliminary (unpublished) studies of isometry indicated that when we modelled the tenodesis as a broad band comprising five straight lines arising from Gerdys tubercle, the most isometric femoral sites were distributed in a way requiring the tenodesis to have a 180 twist between the tibial and femoral attachments. The first purpose of this study was to determine if load state significantly affects the femoral location most isometric to the apex of Gerdys tubercle. The second purpose was to determine how the distances for each of the five lines representing the tenodesis changed with flexion angle for the different load states. The third purpose was to test the hypothesis that a 180 twist in the tenodesis would reduce the range of changes in distance (difference between the largest and smallest changes among the lines for a given angle of flexion) between the tibiofemoral attachments compared with the tenodesis without the twist. Such a reduction would enhance isometry among the lines of the tenodesis.
one

Lateral (A) and anterior (B) views of the testing apThe abduction-adduction adjustment rod enabled us paratus. to adjust the line of action of the externally applied flexion moment to pass through the intercondylar eminence of the tibia in the frontal plane.

Figure 1.

MATERIALS AND METHODS

Our methods entailed quantifying the motion of the knee for four different load states, digitizing potential tibial and femoral attachment sites for the iliotibial band tenodesis, and, using this combined information, predicting the most isometric attachment sites on the femur for a given tibial attachment and the corresponding separation distances between the tibial and femoral attachments.
medius tendons were the primary tissues in the clamp. Thus, the load direction was along the lines of action of these muscles. This was an approximation of physiologic quadriceps muscles loading. The tibia was secured vertical to the testing apparatus. By adjusting the length of the cable, the desired angle of the femur, and hence, the desired flexion angle of the knee, was obtained. A gravity-activated goniometer was used to identify the angle of flexion of the knee. For quadriceps muscles loading, a weight was hung from the femoral rod at prescribed distances from the center of the knee joint and in the midfrontal plane of the knee. This was done to produce external flexion moments of 27.9, 35.1, 39.5, 39.0, 38.2, 33.9, 29.5, and 24.3 N-m for flexion angles from 15 to 120, respectively, in increments of 15. For each angle of flexion, quadriceps muscles force was applied to equilibrate the knee for the flexion moment applied. For full extension of the knee, an external flexion moment was not applied. A quadriceps muscles force of 462 N was applied to produce an extension moment of 20.1 N-m at full extension. This extension moment was equilibrated by the flexion moment generated internally by the ligaments connecting the tibia and femur.

Specimens
Fifteen knees with a mean age of 51.9 years (range, 28 to 76) were obtained fresh at autopsy from 15 cadavers. They were transected 20 cm above and below the joint line, frozen at -20C, and thawed in a cold-water bath before use in this study. The knee was stripped of all soft tissues except the joint capsule, ligaments, patellar tendon, and quadriceps tendon. The specimens were kept moist with saline throughout the experiments.

Experimental setup
The motion and bony landmarks and surfaces of the knees were quantified with a three-dimensional digitizer (Polhemus, Inc., Colchester, VT). The digitizing system included an electromagnetic source attached to the femur, a receiver attached to the tibia, and a receiver in a stylus for manually digitizing bony landmarks and surfaces. The knee extension apparatus used was similar to those previously reported (Fig. 1).14,17 Stiff fiberglass-reinforced plastic rods, 2.5 cm in diameter, were secured to the femur and tibia. A hand-operated turnbuckle was used to apply quadriceps muscles force through a cable clamped to the quadriceps tendon. The rectus femoris and vastus inter-

Testing procedure
Four sets of experiments, three for passive loading and one for quadriceps muscles loading, were performed for flexion angles from 120 of knee flexion to full extension. For passive loading, the knee was manually extended in neutral, internal, and external rotation. For internal and external rotation, the knee was rotated to firm end points. The motion of the knee was quantified during the load states. The

188

used to predict distances between the of attachment. The knee was then dissected free of all soft tissues. Bony landmarks on the tibia and femur were selected and digitized to establish the coordinate systems for the tibia and femur. The posterior two thirds of the lateral side of the lateral femoral condyle and adjacent lateral femoral shaft were digitized to determine the distances between the femoral locations and each of the five points digitized on Gerdys tubercle (Fig. 2). Less than 0.5 mm of spacing was
motion data
were

potential points

the insertion of the iliotibial band into the tibia, were then digitized over a straight-line distance of 1.0 cm with 2.5 mm spacing between adjacent points. The straight line was perpendicular to the direction of fiber insertion. The points were numbered consecutively from the medial-most insertion, point one, to the lateral-most, point five (Fig. 3). The middle point was on the apex ofGerdys tubercle. This numbering system was used to identify the line number; lines one through five of the tenodesis originated at iliotibial band insertion points one through five, respectively.

maintained between adjacent points. Next, the center of the femoral insertion of the lateral collateral ligament was digitized. Five points, representing

Figure Region digitized on the lateral surface of the lateral femoral condyle. More than 8000 points were digitized on
each knee.

2.

Figure 3. The numbering system used to identify the lines and their attachments to the tibia and simulated washer on the femur are shown for the iliotibial band screw tenodesis without (A) and with (B) the 180 twist.

189

We attached a 1-cm diameter washer to the femur at Krackows point F9,16 the junction of the lateral femoral condyle and femoral shaft. Five points on the perimeter of the washer were digitized to represent the attachment of Muellers iliotibial band tenodesis2 to the washer. The tenodesis was modelled both with and without a 180 twist. To model the attachment of the tenodesis without the twist, points one through five on the tibia were attached by straight lines to points one through five, respectively, on the perimeter of the washer (Fig. 3A). To model the twist, points one through five on the tibia were attached to points five through one on the femur, respectively (Fig. 3B). Data

nificance. A paired Students t-test was performed to determine whether a 180 twist in the graft would significantly reduce the range of changes in distance among the five lines compared with the graft without the twist. An a 0.006 level of significance was used based on a Bonferroni adjustment for comparisons at eight angles of flexion.
=

RESULTS
Most isometric femoral locations The femoral location most isometric to the apex of Gerdys tubercle was different for the different load states (Table 1) (Fig. 4). The anteroposterior and proximal-distal coordinates of the most isometric femoral location for quadriceps muscles loading were different (P < 0.015) from those found for the neutrally and externally rotated knees and different in the anteroposterior coordinate from that found for the internally rotated knee. The anteroposterior and proximaldistal coordinates of the most isometric femoral locations for the passive loading states were also different (P < 0.015), except for the proximal-distal coordinates, when comparing the neutrally and externally rotated knees. The most isometric femoral regions for the externally rotated and neutrally rotated knees were elliptical in shape (Fig. 5, A and B). The major axes of the elliptical areas were oriented primarily in a proximal to distal direction. The least changes in tibiofemoral separation distance were found for these loading states (Fig. 6). The most isometric regions for the knees internally rotated and quadriceps muscles loaded were bandlike and oriented primarily in a proximal to distal direction (Fig. 5, C and D). The most or nearly most isometric region for the quadriceps muscles-loaded knee included Krackows point F9 in the over-the-top region of the lateral femoral condyle. The mean changes in tibiofemoral separation distance for the five lines were approximately 8% to 9% for the internally rotated knee and 12% to 14% for the quadriceps muscles-loaded knee (Fig. 6).

analysis

For each load state the motion of the tibia with respect to the femur was recorded from the source attached to the femur and the receiver attached to the tibia. The bony landmarks, which were manually digitized, enabled us to establish coordinate systems for the tibia and femur. This 13 was done so that reported techniques could be used to determine the three-dimensional joint positions during the load states. All of the points digitized on the tibia and femur were transformed to the femoral coordinate system for analysis. Using these data, the distances between each of the potential points of attachment digitized on the lateral side of the lateral femoral condyle and the five points on Gerdys tubercle were predicted. The most isometric point was defined as the femoral location resulting in the least change in distance. Isometry maps were generated to provide visual representations of the variations in distances between the apex of Gerdys tubercle and all femoral sites digitized. Tests were performed to determine the accuracy of our measurements (Appendix 1).
.

Statistical
We

analysis

one-way analysis of variance with reto determine if load state significantly altered the femoral location most isometric to the apex of

performed a peated measures

Gerdys tubercle. One-way analysis of variance with repeated measures was also used to determine if there were significant differences in tibiofemoral separation distance
function of flexion of the knee; this was done for the tenodesis with and angle without the 180 twist and for the quadriceps muscles and external rotation load states. Tukeys &dquo;honest significant difference&dquo; tests were used to perform multiple pair0.05 level of sigwise comparisons at the overall a
among the five lines of the tenodesis
as a
=

Tenodesis attachment to

washer at Krackows

point

F9

For the tenodesis without the twist in the quadriceps muscles-loaded knee, the maximal change in separation distance was 22% at 60 of flexion (Fig. 7A). The distance changes for the outermost lines of the tenodesis (lines one and five) were different (P < 0.01) for angles between 15 and 120. The range of changes among the five lines increased with flexion, reaching a maximum of 19% at 120.

TABLE 1 Coordinates of the most isometric femoral locations for the loads

applieda

Coordinates

are

with respect to the femoral insertion of the lateral collateral ligament. Positive y and

z are

anterior and proximal,

respectively.

190

lines were not significantly different for angles between 60 and 90 (Fig. 8B). The twist reduced the range of changes in separation distance (P < 0.001) for flexion angles between 15 and 120. The maximal reduction was 61% at 15

(Table 3).

DISCUSSION
Two limitations of this study should be considered when interpreting the results. First, the iliotibial band tenodesis was modelled as a broad band comprising five straight lines. In actuality, the tenodesis would have curved over the lateral femoral condyle when attached to Krackows point F9 .16 However, this curvature is in a plane perpendicular to the flexion-extension plane. A planar model demonstrated that the changes in separation distance were primarily determined by motion of the knee in the flexionextension plane (Appendix 1). Thus, the straight-line approximation would be expected to have only a small effect on the results of our study. Second, we did not consider the restraining force in the tenodesis. Thus, the relationship between tibiofemoral separation distance and restraining force was unknown. We are reporting on changes in separation distance between potential tibiofemoral attachment sites and not on the changes in length or strain of the fibers of the tenodesis. This is because we did not measure the length of the fibers of an actual tenodesis during the load states. However, in parts of the following discussion inferences about the strain of a fiber were made based on the assumption that the fiber would be under tension throughout the range of motion of the knee. The femoral location most isometric to Gerdys tubercle was significantly changed by the loading state. Thus, the changes in distance between the tibial and femoral attachment sites of the tenodesis were different for the different loading states. We compared our results with those of Sidles et al.24 where comparisons could be made. In agreement with Sidles et al., there were no perfectly isometric sites. When modelling the tenodesis as a single line arising from the apex of Gerdys tubercle, we found the most isometric location on the femur for the neutrally rotated knee to be close, 4.4 mm, to that reported by Sidles et al. However, our results for the most isometric location during quadriceps muscles loading were quite different from Sidles et al. They found the most isometric location for quadriceps muscles loading to be approximately 10 mm proximal and 6 mm posterior to the insertion of the lateral collateral ligament, whereas our location was approximately 4 mm distal and 10 mm posterior to the insertion of the lateral collateral ligament. Also, the band of isometry representing the least changes in separation distance for the quadriceps muscles-loaded knee in the present study often included the over-the-top region, contrary to the results of Sidles et al. We attribute these differences primarily to the differences in the magnitudes of quadriceps muscles forces used. They applied very low levels; those in this study approached levels generated during stair climbing. These were equivalent to one third the maximal voluntary isometric quadriceps muscles forces reported for a healthy group of adults. 18,26

Lateral view of the most isometric points for quadpassive loading. The origin of the Y and Z axes represents the insertion of the lateral collateral ligament. Positive Y is directed anteriorly and positive Z is directed proximally. The lengths of the Y and Z axes each represent 1 cm. PE, passive extension with external rotation; PN, passive extension with neutral rotation; PI, passive extension with internal rotation; Q, extending the knee with quadriceps muscles forces.

Figure 4.

nceps and

For the tenodesis with the twist in the quadriceps muscles-loaded knee, the changes in separation distance for the outermost lines were not significantly different for angles between 60 and 90 (Fig. 7B). The twist reduced the range of changes in separation distance (P < 0.001) for flexion angles between 60 and 120. The maximal reduction was 44% at 75 of flexion (Table 2). For the tenodesis without the twist in the externally rotated knee, the maximal change in separation distance was -26% at 120 (Fig. 8A). The changes for the outermost lines of the graft (lines one and five) were different (P < 0.01) for angles between 15 and 120. The range of changes among the five lines increased with flexion, reaching a maximum of 26% at 120. For the tenodesis with the twist in the externally rotated knee, the changes in separation distance for the outermost

191

Figure 5. Typical results of computer-generated isometry maps for a single knee for the externally rotated (A), neutrally rotated (B), internally rotated (C), and quadriceps muscles-loaded knees (D). The maps represent percentage changes in tibiofemoral separation distance for a single point-to-point attachment arising from the apex of Gerdys tubercle.
For the quadriceps muscles-loaded knee, the changes in tibiofemoral separation distance for the tenodesis without the twist and attached to the simulated washer at Krackows point F9 were 22% for lines 1 and 2 on the anterior edge of the graft. These changes in distance would be expected to represent those occurring during quadriceps muscles loads when initial tensioning of the tenodesis is performed to remove the slack from the tenodesis (i.e., with no slack at any angle of flexion). For the externally rotated knee, the changes in distance were more than 26% for line 5 on the posterior edge of the tenodesis. The changes in distance found for the externally rotated knee would be

192

Figure 6. Normalized changes in tibiofemoral separation distance for the five lines of the tenodesis attaching to their most isometric points. In this and the following figures f1 through f5 are fiber numbers.

expected to represent those occurring during quadriceps muscles loads with the tenodesis attached to the simulated washer at Krackows point F9 but pulled tight at surgery to externally rotate the tibia. The 180 twist reduced the maximal amount of separation distance found for the quadriceps muscles-loaded knee to 20% and for the externally rotated knee to 18%. These strains would be more than the strain (i.e., 14.5%) that has been found to lead to failure of the fascia latae. However, it is important to understand that these changes in distance depend on the load state and, thus, the knee motion generated. Whether these high levels of strain would occur physiologically depends on the strength of the graft. A graft of sufficient strength could alter the motion of the knee enough to reduce the changes in tibiofemoral separation distance and, thus, reduce the strains in the graft to physiologically acceptable levels. However, this study does not address the strength of the graft.
The incorporation of a 180 twist in the tenodesis was found to reduce significantly the range of changes in separation distance. This should enhance load sharing among the fibers because strain and tension in an elastic material are directly proportional according to ~ T/A Ee, where ~ is stress, T is tension, A is area, E is Youngs modulus of elasticity, and e is strain. Assuming all of the fibers have the same elasticity and cross-sectional areas, the tensions in the fibers across the breadth of the tenodesis become more similar as the strains become more similar.
= =

Figure 7. Normalized changes in separation distance during quadriceps muscles loading for the tenodesis attached to a simulated washer at Krackows point F9 without the twist (A) and with the 180 twist (B), N 13.
=

We have previously demonstrated, in a cadaveric study, that the iliotibial band tenodesis was incapable of returning normal anterior laxity to the ACL-deficient knee.However, in vitro studies performed in our laboratory and

193 TABLE 2

Ranges of changes in length among the fibers of the for the quadriceps muscles-loaded knee

tenodesis

Data are normalized with respect to 0 of knee flexion (see text

for

equation).

b Percentage difference equals (With twist-Without twist)/ Without twist. Paired Students t-test.
C

TABLE 3

Ranges of changes in length among the fibers of the tenodesis for the passively extended and externally rotated knee

a Data are normalized with respect to 0 of knee flexion (see text for equation). All differences are significant at P < 0.001. b Percentage difference equals (With twist-Without twist)/ Without twist.

tenodesis provided support to the healing intraarticular graft. Our results indicate that the incorporation of the 180 twist should decrease the range of strain among the fibers across the breadth of the tenodesis, thereby enhancing isometry among the fibers of the tenodesis. This implies that a 180 twist in the tenodesis would enhance load sharing among the fibers of the tenodesis and, therefore, would enhance the overall strength of the tenodesis.

ACKNOWLEDGMENTS
was supported by Grant AR40605 from the National Institute of Arthritis & Musculoskeletal Skin Diseases. We gratefully acknowledge the assistance of Theodore Karrison, PhD, Department of Medicine, University of Chicago, for advice on statistical treatment.

This work

Figure 8. Normalized changes in separation distance during external rotation for the tenodesis attached to a simulated washer at Krackows point F9 without the twist (A) and with the 180 twist (B), N = 13.
other laboratories 1 have demonstrated that the extraarticular graft can share the load with an intraarticular graft. Furthermore, based on the results of their clinical study, Noyes and Barber&dquo; hypothesized that the iliotibial band

REFERENCES
Andriacchi TP, Andersson GB, Fermier RW, et al: A study of lower-limb mechanics during stair-climbing J Bone Joint Surg 62A 749-757, 1980 2 Arnold JA, Coker TP, Heaton LM, et al Natural history of anterior cruciate tears Am J Sports Med 7 305-313, 1979 3 Bertoia JT, Urovitz EP, Richards RR, et al Anterior cruciate reconstruction using the MacIntosh lateral-substitution over-the-top repair. J Bone Joint Surg 67A 1183-1188, 1985 4 Butler DL, Grood ES, Noyes FR, et al: Effects of structure and strain measurement technique on the material properties of young human tendons and
1

194
fascia J Biomech 17 579-596, 1984 5. Carson WG Jr Extra-articular reconstruction of the anterior cruciate ligament Lateral procedures Orthop Clin North Am 16 191-211, 1985 6. Clancy WG Jr, Nelson DA, Reider B, et al. Anterior cruciate ligament reconstruction using one-third of the patellar ligament augmented by extraarticular tendon transfers J Bone Joint Surg 64A: 352-359, 1982 7. Draganich LF, Reider B, Ling M, et al An in vitro study of an intraarticular and extraarticular reconstruction in the anterior cruciate ligament deficient knee Am J Sports Med 18. 262-266, 1990 8. Draganich LF, Reider B, Miller PR: An in vitro study of the Muller anterolateral femorotibial ligament tenodesis in the anterior cruciate ligament deficient knee. Am J Sports Med 17. 357-362, 1989 9 Drez D Jr Modified Enksson procedure for chronic anterior cruciate in30-36, stability. Orthopedics1 1978 10 Engbretsen L, Lew WD, Lewis JL, et al The effect of an iliotibial tenodesis on intraarticular graft forces and knee joint motion Am J Sports Med 18. 16. Krackow KA, Brooks RL: Optimization of knee ligament position for lateral extraarticular reconstruction Am J Sports Med 11 293-302, 1983 17. Kurosawa H, Walker PS, Abe S, et al Geometry and motion of the knee for implant and orthotic design J Biomech 18 487-499, 1985 18. Lieb FJ, Perry J: Quadnceps function. An electromyographic study under isometric conditions J Bone Joint Surg 53A: 749-758, 1971 19 Lipscomb AB, Woods GW, Jones A. A biomechanical evaluation of the iliotibial tract screw tenodesis. Am J Sports Med 20: 1992 742-745, 20 Mueller W The Knee. Form, Function, and Ligament Reconstruction New 253-257 York, Springer-Verlag, 1983, pp 21 Noyes FR, Barber SD: The effect of an extra-articular procedure on allograft reconstructions for chronic ruptures of the anterior cruciate ligament J Bone Joint Surg 73A: 882-892, 1991 22. Puddu G: Model for reconstruction of the anterior cruciate ligament using the semitendinosus tendon. Am J Sports Med 8 402, 1980 23. Shelbourne KD, Nitz P: Accelerated rehabilitation after anterior cruciate ligament reconstruction. Am J Sports Med 18 292-299, 1990 24 Sidles JA, Larson RV, Garbini JL, et al Ligament length relationships in the moving knee J Orthop Res 6: 593-610, 1988 25. Simonet WT, Sim FH: Repair and reconstruction of rotatory instability of the knee. Am J Sports Med 12 89-97, 1984 26. Smidt GL Biomechanical analysis of knee flexion and extension J Biomech
27.
et al How iliotibial tract Injuries of the knee combine with acute anterior cruciate ligament tears to influence abnormal anterior tibial displacement. Am J Sports Med 21. 1993 55-60, 28. Wroble RR, Grood ES, Cummings JS, et al: The role of the lateral extraarticular restraints in the anterior cruciate ligament-deficient knee Am J Sports Med 21: 257-262, 1993

169-176, 1990 JM, Blazina ME, Del Pizzo W, et al: Extra-articular stabilization of the knee joint for anterior instability. Clin Orthop 147 56-61, 1980 12. Fried JA, Bergfeld JA, Weiker G, et al. Anterior cruciate reconstruction using the Jones-Ellison procedure J Bone Joint Surg 67A: 1029-1033,
11. Fox

1985 13. Grood ES, Suntay WJ: A joint coordinate system for the clinical description of three-dimensional motions Application to the knee J Biomech 105

136-144, 1983 14 Harding ML, Harding L, Goodfellow JW. A preliminary report of a simple rig to aid study of the functional anatomy of the cadaver human knee joint. J Biomech 10 517-523, 1977 15. Jensen JE, Slocum DB, Larson RL, et al: Reconstructive procedure for anterior cruciate ligament insufficiency A computer analysis of clinical results. Am J Sports Med 11: 240, 1983

6: 79-92, 1973 Terry GC, Norwood LA, Hughston JC,

APPENDIX 1

Computation of distance changes

The distance change, defined as D (DmaX-Dm~n)/DmaX, was used to quantify the distance between each tibial and femoral attachment site pair. DmaX was the maximal distance and Do was the minimal distance found for the entire range of motion, 0 to 120 of flexion. Distance was determined using the Pythagorean theorem in three dimensions. The most isometric point was defined as the femoral location resulting in the least change in distance. The normalized distance between the tibia and femur was computed as percentage D~ =100 x (D,-Do)/Do, where D, is the distance for flexion angle i, and Do is the distance for full extension. Normalized distance was plotted in Figures 6, 7, 8, and A2.
=

calipers accurate to 0.025 mm. An area of 1.0 cm2 surrounding the insertion of the iliotibial band substitute on the femur was digitized. The insertion of the suture into the tibia was also digitized. We found the length of the iliotibial band
suture to be 0.7 mm from both the mean and median of that predicted with our methods over the range of motion. The most isometric femoral point predicted for the iliotibial band

Accuracy of measurements
To determine the accuracy of the digitizing system, a plastic plate was etched to produce an array of 40 points over a square area of 203.2 x 203.2 mm (8 x 8 inch). The relative coordinates of the points were known to 0.025 mm. The plate was positioned over the test platform and within the space through which the knee would flex during the experiments. Each of the 40 points were digitized with the stylus. This was repeated five times. The average and maximal root-mean-square errors of the digitized points from their known positions were 0.57 and 1.01 mm, respectively. We used a plastic knee linked by stiff sutures representing the iliotibial band tenodesis and the cruciate ligaments to assess the accuracy of our measurement system. The source was mounted on the femur and the receiver was mounted on the tibia, closely simulating their placements on the cadaveric knees. With the sutures taut to represent isometricity, we flexed the knee from full extension to 120 of flexion. The length of the taut iliotibial band suture was measured with dial

Figure A1. Illustration of the planar model. The circle, representing the femur, rotates counterclockwise, representing flexion of the knee. The base represents the stationary tibia. f1, f3, and f5 represent the anterior, middle, and posterior
fibers of the simulated tenodesis.

195

Figure A2. Normalized changes

in

separation distance for the planar model without the

twist

(A)

and with the 180 twist

(B).

suture was within 1.0 mm of the actual insertion of the suture. These findings are similar to those reported earlier for
a

similar

experimental setup using the Polhemus digitizing

system.24
Planar model

A planar model of tibiofemoral motion during flexion of the knee was developed to explain the reduction in the range of changes in separation distance between the tibia and femur that occurred with a 180 twist in the graft (Fig. A1 ). We modelled a graft 1 cm wide with three straight lines to

represent the fibers on either edge and in the middle of the graft. The tibia was stationary. The femur was modelled as a circle of 8 cm diameter that rotated in a plane about its center. The attachments of the graft to the tibia and femur were over straight line distances of 1 cm each. The twist substantially reduced the range of changes in distance (difference between the largest and smallest changes in distance among the lines) for a given angle of flexion, thereby enhancing isometry among the lines (Fig. A2). Furthermore, the model demonstrated that the reduction was due to (single degree-of-freedom) rotation about the flexion-

extension axis of the knee.