A Biomechanical Comparison of the EndoButton CL Using Transtibial Drilling and EndoButton Direct Using Anteromedial Arthroscopic Drilling Chealon D. Miller, M.D., Andrew C. Gerdeman, M.D., Chase G. Bennett, B.S., Joseph M. Hart, Ph.D., and Mark D. Miller, M.D.
Purpose: To compare the biomechanical properties of the EndoButton Direct (Smith & Nephew, Andover, MA) with the EndoButton CL (Smith & Nephew) using 2 different drilling techniques. Only the femoral side (and not the tibial side) of the graft fixation complex (bone–fixation device– anterior cruciate ligament [ACL] graft) was examined in this study. Methods: ACL reconstructions were performed on 20 cadaveric knees (10 matched pairs), with an age range from 73 to 89 years, by use of a doubled semitendinosus and gracilis tendon graft. Ten knees underwent femoral tunnel drilling from a standard anteromedial arthroscopic portal, and the EndoButton Direct was used for fixation. Ten knees underwent femoral drilling through a medial transtibial approach, and the EndoButton CL was used for fixation. All graft fixation complexes were subjected to 1,000 loading cycles. Graft elongation after 1,000 cycles, stiffness, ultimate load, and mode of failure were determined for each specimen. Results: The mean failure load was significantly higher for the EndoButton CL (959.9 ⫾ 190.4 N) compared with the EndoButton Direct (697.7 ⫾ 341.8 N) (P ⫽ .05). There was no significant difference in overall stiffness or graft elongation after 1,000 cycles between the 2 fixation devices. Conclusions: The maximum load during ultimate failure testing was higher for the EndoButton CL with transtibial drilling when compared with the EndoButton Direct with anteromedial drilling. There were no differences found between the EndoButton Direct and EndoButton CL with regard to overall stiffness or elongation after cyclic loading. Clinical Relevance: Reduced ultimate failure strength has implications in reconstructed patients if forces imparted on the ACL exceed the strength of graft fixation.
A
nterior cruciate ligament (ACL) reconstruction has been largely successful, but controversies remain regarding surgical technique.1 Femoral tunnel placement and single- versus double-bundle reconstruction remains a subject of intense debate.2-4 Al-
From the Department of Orthopaedic Surgery, University of Virginia, Charlottesville, Virginia, U.S.A. Supported by Smith & Nephew, Andover, MA. The authors report no conflict of interest. Received September 26, 2009; accepted February 12, 2010. Address correspondence and reprint requests to Mark D. Miller, M.D., Department of Orthopaedic Surgery, University of Virginia, 400 Ray C Hunt Dr, Ste 330, Charlottesville, VA 22908, U.S.A. E-mail:
[email protected] © 2010 by the Arthroscopy Association of North America 0749-8063/9566/$36.00 doi:10.1016/j.arthro.2010.02.018
though the location of the intra-articular tunnel entrance is the most important factor for establishing rotator stability, some surgeons advocate more lateral graft placement of the reconstructed ACL within the femoral notch to improve rotatory stability.5-8 More lateral tunnels have been shown to lead to shorter femoral tunnels with subsequently less graft within tunnels.2 The EndoButton Direct (Smith & Nephew, Andover, MA) was introduced in mid 2007 to address some of the concerns about shorter femoral tunnels. The EndoButton Direct attaches graft directly to the EndoButton as opposed to its predecessor, the EndoButton CL (Smith & Nephew), which uses a loop of polyester tape to indirectly link the graft to the EndoButton. Femoral tunnels placed at relatively more horizontal orientations in a flexed knee joint usually result in shorter femoral tun-
Arthroscopy: The Journal of Arthroscopic and Related Surgery, Vol 26, No 10 (October), 2010: pp 1311-1317
1311
1312
C. D. MILLER ET AL.
nels. Previous studies have shown that the mean length of these more horizontal tunnels is between 23 and 37 mm depending on exact placement.2 Given that the shortest tape length on the EndoButton CL is 15 mm, the tape may occupy a significant amount of tunnel length, resulting in tunnels insufficient in length and volume for graft incorporation. Use of the EndoButton Direct maximizes the amount of graft in the tunnel and may increase the stiffness of the whole complex, thus minimizing graft movement within the tunnel. However, to our knowledge, there have been no biomechanical studies on the EndoButton Direct that have been published. The purpose of this study is to compare the fixation properties of the EndoButton Direct (anteromedial drilling) with the EndoButton CL (transtibial drilling) by means of biomechanical analysis of human cadaveric knees. We hypothesize that the EndoButton CL will have a lower stiffness than the EndoButton Direct because of the placement of the polyester loop in series with the graft. We expect there to be no difference in ultimate load between the EndoButton CL and the EndoButton Direct. Only the femoral side (and not the tibial side) of the graft fixation complex (GFC) (bone–fixation device–ACL graft) was examined in this study. METHODS We used 10 matched pairs (N ⫽ 20) of cadaveric knees (midshaft femur distally to midshaft tibia; age range, 73-89 years). Each matched pair of knee specimens came from the same donor. The knees from each pair were randomly assigned by a random number generator, where one of the knees from each pair
FIGURE 1.
(right or left) was randomly assigned to the anteromedial group (EndoButton Direct) and the other to the transtibial group (EndoButton CL). This resulted in 10 limbs (5 right and 5 left) in the anteromedial group and 10 limbs (5 right and 5 left) in the transtibial group. In the transtibial group, a femoral tunnel was drilled arthroscopically from a medial location on the tibia, immediately anterior to the fibers of the superficial medial collateral ligament. In the anteromedial group, a femoral tunnel was drilled arthroscopically from a standard anteromedial arthroscopic portal used for knee arthroscopy. The intra-articular location for all tunnels was chosen based on established borders—in the posteromedial footprint of the ACL, lateral to the posterior border of the anterior horn of the lateral meniscus, adjacent to the upslope of the medial tibial eminence, and 7 mm anterior to the crossing fibers of the posterior cruciate ligament.9 Guide pin placement was performed with a 55° guide (Acufex; Smith & Nephew). After guide pin placement, the tunnels were all drilled with a 10-mm-diameter cylindrical drill (Smith & Nephew).The specimens that were drilled by transtibial drilling received an EndoButton CL, whereas those drilled from the anteromedial arthroscopic portal received the EndoButton Direct (Fig 1). Each cadaver was stored at ⫺20°C before being thawed for use. Each specimen was thawed at least 24 hours before harvesting of the hamstring grafts. The hamstring grafts were harvested by a fellowshiptrained orthopaedic surgeon using a 5-cm oblique incision on the medial surface of the tibia. A tendon stripper was used to harvest the semitendinosus and gracilis hamstring graft for each cadaveric specimen.
(A) EndoButton Direct and (B) EndoButton CL.
DRILLING TECHNIQUES FOR ENDOBUTTON After harvesting, each graft was prepared by use of a commercially available graft preparation tray (Graftmaster 2000; Arthrex). We whip-stitched a semitendinosus graft for 8 throws on both ends using a No. 2 nonabsorbable braided suture and performed the same procedure for the gracilis tendon graft. We then combined the grafts to create a quadruple hamstring graft that was looped over in the middle. We then placed the implant (EndoButton CL or EndoButton Direct) at the middle portion of the quadrupled hamstring graft for placement in the cadaveric specimens. After harvesting of the grafts, each femur had its soft tissue removed and was disarticulated from the tibia. We then placed the EndoButton CL with graft in the femurs drilled through a transtibial approach and the EndoButton Direct with graft in the femurs drilled through the anteromedial arthroscopic portal. We only used 15-mm EndoButton CL for this study. Each free end of the graft was separately and equally tensioned to prepare for biomechanical testing. Each femur was secured by use of epoxy resin (Bondo; 3M, St. Paul, MN) in a 3-inch-diameter piece of polymerized vinyl chloride (PVC) piping that accommodated the entire length of the femur without compromising the implant. Each potted femur specimen was allowed to harden for 30 minutes before biomechanical testing. One specimen was discarded from all testing because of posterior wall blowout during graft passage with the EndoButton Direct. Biomechanical Testing and Analysis To test our hypothesis, after the preparation of the specimens, they were mounted on a structural testing frame (Bionix; MTS, Eden Prairie, MN). Inclusion criteria for biomechanical testing included the presence of hamstring autograft and no tunnel blowout during graft placement. The femur was fixed to the endplate of the load cell by use of a 1-cm-thick aluminum U-bracket and 2 transverse lag screws (Fig 2). The suture end of the graft was then clamped between 2 plates approximately 1 cm distal to the medial condyle so that the graft was enclosed within the clamp. The surface of the clamp plates was grated to prevent graft slippage. When fixing the femur to the frame, special care was taken to orient the long axis of the tunnel in line with the applied tension. Orientation was performed by direct visualization of the tunnel exit and entrance, and the femur was rotated within the apparatus until the graft was in a straight line with the line of pull of the testing apparatus. This orienta-
1313
FIGURE 2. Apparatus used for biomechanical testing showing femur potted in PVC pipe, and MTS machine orientation.
tion maximized load on the fixation by minimizing friction between the tunnel and the graft. At this orientation, there was approximately 2 cm between the intra-articular tunnel opening and the highest point of the medial condyle. The suture end of the graft was then clamped between 2 plates approximately 1 cm distal to the medial condyle so that the graft was enclosed within the clamp. The distance between the intra-articular tunnel opening and the clamp site was approximately 3 cm for each specimen, which simulates the intra-articular length of the ACL.10 These measurements were done with a ruler just before testing. Cyclic Loading Each specimen was then preloaded to 10 N and subjected to 1,000 loading cycles between 10 and 150 N at a frequency of 30 cycles/min.11 During testing, the graft was manually hydrated every 5 minutes with saline solution to prevent desiccation. The GFC was defined as the graft, fixation device, and osseous surface in which the graft and fixation device rested. Mean values (⫾standard deviation) were calculated for GFC elongation (in millimeters) by subtracting the initial GFC length during the 10-N preload from the final GFC length at 10 N after cyclic loading.12 Displacement was measured continuously by the MTS
1314
C. D. MILLER ET AL.
system during testing. These values were used for the GFC elongation calculations. Load to Failure After completion of the cyclic loading test, each specimen was allowed to rest for 20 minutes under 10 N of tension while the graft was rehydrated every 5 minutes. After the recovery period, the specimens were axially loaded in tension to failure at a constant rate of 200 mm/min. The ultimate load (in newtons) and overall stiffness (in newtons per millimeter) were recorded for each specimen. Data from the MTS frame were imported into a spreadsheet (Excel; Microsoft, Seattle, WA) and plotted. Two of the specimens had ultimate loads very late in elongation (⬎15 mm) that occurred after a definite local maximum had occurred on the force-displacement curve (Fig 3). For these 2 specimens, we decided to use the first local maximum obtained, which was defined as the maximum load occurring before a 10% drop of that load occurred. This definition was used as a way to isolate the maximum physiologically relevant load in these cases and is unique to this study. Only the first ultimate load is clinically significant because a greater elongation is needed to reach later peaks. Overall stiffness was defined as the slope between the 10-N resting period immediately preceding the failure test and the ultimate load. Mode of Failure After detachment from the frame, the specimens were visually inspected for mode of failure and classified as having midsubstance rupture of the graft, implant migration into the bone, implant failure, or graft pullout from the fixation.11 Midsubstance rupture was defined as graft tearing somewhere between the
clamp and implant, whereas graft pullout was defined as a graft tear at the point of implant attachment. Implant failure denoted damage to the implant itself, and implant migration was defined as a failure of the cortical bone to support the implant, causing it to begin pulling through the bony margins of the tunnel. Statistical Analysis We used independent-sample t tests to compare elongation after cyclic loading between the 2 fixation devices and their corresponding drilling techniques. Data are presented as mean ⫾ standard deviation unless otherwise specified. We also used independentsample t tests to compare overall stiffness and ultimate failure strength between the 2 fixation devices and drilling techniques. The Pearson 2 test was used to compare the mechanisms of failure. Any test was considered statistically significant if the associated P value was .05 or less. SPSS software (version 17.0; SPSS, Chicago, IL) was used for all statistical tests. RESULTS One specimen from the anteromedial group (EndoButton Direct) was discarded from all testing because of graft tunnel destruction during passage of the GFC. Cyclic Loading In our initial test protocol, we clamped the MTS actuator to the suture leads from the whipstitch instead of to the graft itself. This resulted in abnormal elongation data, presumably from suture tightening within the graft. We tested 2 EndoButton CL specimens and 1 EndoButton Direct specimen under this protocol, and they were not included in the cyclic loading data.
FIGURE 3. Representative results from ultimate failure testing, with EndoButton CL showing ultimate load near 1,000 N (left) and EndoButton Direct showing ultimate load near 600 N (right).
DRILLING TECHNIQUES FOR ENDOBUTTON One of the EndoButton Direct specimens pulled through the tunnel 100 cycles into cyclic loading and was not included in the data series for GFC elongation. The final data used for GFC elongation included 7 EndoButton Direct specimens and 9 EndoButton CL specimens. There were no significant differences observed in GFC elongation after cyclic loading for the EndoButton Direct group (4.5 ⫾ 1.9 mm) versus the EndoButton CL group (3.9 ⫾ 1.0 mm) (Table 1). Load to Failure and Stiffness The mean failure load was significantly higher for the EndoButton CL (959.9 ⫾ 190.4 N) compared with the EndoButton Direct (697.7 ⫾ 341.8 N) (t17 ⫽ 2.1, P ⫽ .05). There were no statistically significant differences in overall stiffness measured during load-tofailure testing between the EndoButton Direct (80.9 ⫾ 18.8 N/mm) and EndoButton CL (75.6 ⫾ 14.4 N/mm). Mode of Failure There was no significant difference between fixation devices for the mode of failure during load-tofailure testing (2 ⫽ 4.0, P ⫽ .14). Failure occurred because of implant migration in 5 of 10 EndoButton CL specimens and 7 of 9 EndoButton Direct specimens. Midsubstance rupture accounted for the other 5 EndoButton CL failures. The other 2 EndoButton Direct specimens failed by either graft pullout or by
TABLE 1.
Biomechanical Tests Performed EndoButton Direct
EndoButton CL
Ultimate tensile strength (load-to-failure test) (mean ⫾ SD) (N) 697.7 ⫾ 341.8 959.9 ⫾ 190.4 Linear stiffness during load-to-failure test (mean ⫾ SD) (N/ mm) 80.9 ⫾ 18.8 75.6 ⫾ 14.4 Elongation during cyclic loading (mean ⫾ SD) (mm) 4.5 ⫾ 1.9 3.9 ⫾ 1.0 Mode of failure (No. of specimens) Implant migration into bone 7 5 Implant failure 0 0 Graft pullout from fixation 1 0 Midsubstance rupture of graft 1 5
P Value
.051
.51
.46 .14
1315
midsubstance rupture. No specimens in either group failed because of breakage of the implant itself (Table 1). DISCUSSION We observed significantly lower loads at the point of ultimate failure during elongation of the GFC when using the EndoButton Direct, as compared with the EndoButton CL. Of 9 specimens tested with the EndoButton Direct, 3 failed at loads less than 400 N, and another failed at 455 N. The lowest load to failure in the EndoButton CL group occurred at 654 N. Because the ages of the specimens ranged from 70 to 89 years, this may have caused early failure in the EndoButton Direct group; however, we attempted to control for the possibility of weaker specimens in any particular group by having a matched pair in the opposite group. The failure load of the EndoButton Direct is of particular concern during early postoperative physical therapy and ambulation. Shelburne et al.13 report that the ACL experiences loads up to 303 N during level walking. Lu and Lu14 estimate that the ACL experiences loads up to 52% of the patient’s body weight (324 N for a 140-lb individual) during stair descent. The loads predicted in these studies come within 200 N of the failure loads found for 4 of the EndoButton Direct specimens. We found 2 possible reasons for the difference in strength between the 2 fixation methods. One possible explanation is the design and structure of the EndoButton Direct. It has an open end and a closed circular end (Fig 1A). The smaller cortical footprint on the open (pronged) end of the EndoButton Direct likely resulted in a higher force/area (pressure) under equal loading conditions on its open end in comparison to the EndoButton CL, which does not have an open end (Fig 1). The higher pressure exerted by the open end of the EndoButton Direct is compounded by the fact that the extra-articular femoral surface of the anteromedial arthroscopic tunnel lies closer to the joint, placing it on metaphyseal bone rather than the diaphyseal site on which the EndoButton CL rests. Metaphyseal bone has been shown to have weaker biomechanical properties compared with diaphyseal bone; therefore placement of the EndoButton Direct on metaphyseal bone may explain the lower loads to failure.15 We believe that the combination of these factors allowed the open end of the EndoButton Direct to pull through the cortex at lower loads.
1316
C. D. MILLER ET AL.
Another possible explanation for the findings of earlier EndoButton Direct failure is that to accommodate the EndoButton Direct, a larger-diameter tunnel must be created at the external surface of the femur. This has the effect of creating a tunnel that (1) is a larger percentage of the implant length for the EndoButton Direct and (2) creates a larger osseous breach in an area of relatively more compromised bone (metaphyseal v cortical). Therefore the EndoButton Direct is more likely to migrate into the tunnel if it shifts postoperatively through a portion of bone that has been compromised because of its size. This is exacerbated by the complex geometry of the bone at the external surface of the anteromedial tunnel. The bone surface may be more oblique to the tunnel axis, leading to a larger tunnel opening, and the curvature of the resting area may prevent the EndoButton Direct from sitting flush on the cortex. Such a scenario would create less cortical area to which the implant could apply load, leading to stress concentrations and a lower failure load. The other 2 structural parameters that our study looked at were graft elongation and overall stiffness. No differences were found between the 2 fixation devices with regard to overall stiffness or GFC elongation. This was contradictory to our original hypothesis that because of the EndoButton CL’s loop placement in series with the graft, any elasticity in the tape would contribute to overall displacement and would thus decrease stiffness. However, this was not observed in our study but may represent a more stiff polyester tape or the inability of small sample sizes to detect a moderate difference in elongation and overall stiffness. We identified several limitations of this study. It was carried out in a cadaveric model, which cannot directly mimic physiologic bone quality. However, the model does allow for a comparative situation in terms of tunnel location. We also only used the 8-mm EndoButton Direct and did not vary the implant size based on patient parameters. Using patient-specific sizing may alter the way the implant sits on the bone and change some of the properties of the fixation, and this should be addressed in future studies. Another limitation is related to our application of load to the GFC. Tension was applied to the graft coaxially to the tunnel, producing a purely tensile force in the GFC. This differs from clinical scenarios in which the maximum traction in the ACL is generated near full knee extension. In this position the femoral tunnel is no longer in line with the graft as it travels through the joint space. We also could have ex-
panded the study to look at 4 groups: EndoButton CL with transtibial drilling, EndoButton Direct with anteromedial drilling, EndoButton Direct with transtibial drilling, and EndoButton CL with anteromedial drilling. However, we believe that our study best represents the clinical scenario in which the respective EndoButton products are most widely used. This study did not include graft preconditioning. This step could have been included in our study, and its omission may have resulted in lower stiffness data and greater GFC elongation.16 This step is carried out both in the clinical arena and in some biomechanical testing protocols.17 The majority of our protocol was based on the study of Milano et al.,12 who did not use graft preconditioning. On more extensive literature review, this would have been an appropriate step to add. CONCLUSIONS The maximum load during ultimate failure testing was higher for the EndoButton CL with transtibial drilling when compared with the EndoButton Direct with anteromedial drilling. There were no differences found between the EndoButton Direct and EndoButton CL with regard to overall stiffness or elongation after cyclic loading. Acknowledgment: The authors acknowledge Olugbenga “Tumi” Oredein for his assistance during data collection and analysis.
REFERENCES 1. Lubowitz JH, Poehling GG. Techniques in double-bundle anterior cruciate ligament reconstruction: As simple as ABC, or putting the cart before the horse? Arthroscopy 2008;24:10891091. 2. Golish SR, Baumfeld JA, Schoderbek RJ, Miller MD. The effect of femoral tunnel starting position on tunnel length in anterior cruciate ligament reconstruction: A cadaveric study. Arthroscopy 2007;23:1187-1192. 3. Amiel D, Kleiner JB, Roux RD, Harwood FL, Akeson WH. The phenomenon of “ligamentization”: Anterior cruciate ligament reconstruction with autogenous patellar tendon. J Orthop Res 1986;4:162-172. 4. Yagi M, Kuroda R, Nagamune K, Yoshiya S, Kurosaka M. Double-bundle ACL reconstruction can improve rotational stability. Clin Orthop Relat Res 2007;454:100-107. 5. Kanamori A, Woo SL, Ma CB, et al. The forces in the anterior cruciate ligament and knee kinematics during a simulated pivot shift test: A human cadaveric study using robotic technology. Arthroscopy 2000;16:633-639. 6. Woo SL, Kanamori A, Zeminski J, Yagi M, Papageorgiou C, Fu FH. The effectiveness of reconstruction of the anterior cruciate ligament with hamstrings and patellar tendon. A ca-
DRILLING TECHNIQUES FOR ENDOBUTTON
7. 8.
9. 10.
11.
daveric study comparing anterior tibial and rotational loads. J Bone Joint Surg Am 2002;84:907-914. Steadman J, Mair S. Anterior cruciate ligament reconstruction with bone-patellar tendon-bone autograft two-incision technique. Oper Tech Orthop 1999;9:273-280. Loh JC, Fukuda Y, Tsuda E, Steadman RJ, Fu FH, Woo SL. Knee stability and graft function following anterior cruciate ligament reconstruction: Comparison between 11 o’clock and 10 o’clock femoral tunnel placement. 2002 Richard O’Connor Award paper. Arthroscopy 2003;19:297-304. Jackson DW, Gasser SI. Tibial tunnel placement in ACL reconstruction. Arthroscopy 1994;10:124-131. Brown CH Jr, Wilson DR, Hecker AT, Ferragamo M. Graftbone motion and tensile properties of hamstring and patellar tendon anterior cruciate ligament femoral graft fixation under cyclic loading. Arthroscopy 2004;20:922-935. Milano G, Mulas PD, Sanna-Passino E, Careddu GM, Ziranu F, Fabbriciani C. Evaluation of bone plug and soft tissue anterior cruciate ligament graft fixation over time using transverse femoral fixation in a sheep model. Arthroscopy 2005;21: 532-539.
1317
12. Milano G, Mulas PD, Ziranu F, Piras S, Manunta A, Fabbriciani C. Comparison between different femoral fixation devices for ACL reconstruction with doubled hamstring tendon graft: A biomechanical analysis. Arthroscopy 2006;22:660668. 13. Shelburne KB, Pandy MG, Anderson FC, Torry MR. Pattern of anterior cruciate ligament force in normal walking. J Biomech 2004;37:797-805. 14. Lu T, Lu C. Forces transmitted in the knee joint during stair ascent and descent. J Mech 2006;22:289-297. 15. Choi K, Kuhn JL, Ciarelli MJ, Goldstein SA. The elastic moduli of human subchondral, trabecular, and cortical bone tissue and the size-dependency of cortical bone modulus. J Biomech 1990;23:1103-1113. 16. Graf BK, Vanderby R Jr, Ulm MJ, Rogalski RP, Thielke RJ. Effect of preconditioning on the viscoelastic response of primate patellar tendon. Arthroscopy 1994;10:90-96. 17. Hoher J, Scheffler SU, Withrow JD, et al. Mechanical behavior of two hamstring graft constructs for reconstruction of the anterior cruciate ligament. J Orthop Res 2000;18:456-461.
