The goal of reconstruction of the posterior cruciate ligament is to restore joint function by recreating as closely as possible normal knee kinematics. Anatomic studies have defined the characteristics of the posterior cruciate ligament and have aided the development of more anatomic reconstructive techniques. Double-bundle reconstruction of the posterior cruciate ligament is believed by some authors to better approximate the anatomy of the posterior cruciate ligament, and several biomechanical studies have shown superior results after this type of reconstruction1,2. Previously, one of us (J.K.S.) and colleagues reported a technique for reconstruction of the posterior cruciate ligament with a bifid bone-patellar tendon-bone allograft3. This graft allows double-bundle reconstruction with both arthroscopic transtibial tunnel and open tibial inlay techniques. It allows use of much larger grafts, which more closely resemble the normal posterior cruciate ligament4,5.
Bergfeld et al. compared single and double-bundle reconstructions of the posterior cruciate ligament in a tibial inlay model6. They found no differences in posterior tibial translation and concluded that it is perhaps unnecessary to perform a double-bundle reconstruction of the posterior cruciate ligament if a tibial inlay construct is utilized. They did not evaluate rotational stability, and the model that they used was an isolated posterior cruciate ligament deficiency model with the posterolateral corner structures remaining intact.
In the first part of this study, we showed how deficiency of the posterior cruciate ligament and posterolateral corner affects clinical and stress radiographic findings in a cadaver model7. In this, second, part of the study, we compared the effects, on the same clinical and radiographic parameters, of single and double-bundle tibial inlay reconstruction of the posterior cruciate ligament with or without an intact posterolateral corner.
Our purpose was to use standard clinical and stress radiographic examinations to compare both posterior tibial translation and external rotation following double and single-bundle tibial inlay reconstruction of the posterior cruciate ligament. Additionally, we aimed to better define, in terms of these same parameters, the role of the posterolateral corner in single and double-bundle reconstructions. We hypothesized that the more anatomic double-bundle reconstruction would restore the parameters seen on clinical and stress radiographic examination more closely to those in the intact condition than would a single-bundle reconstruction, even with a tibial inlay construct.
Specimen and Graft Preparation
Nine of the original ten cadaver knee specimens included in the first part of our study7 were used for this study. The mean age of the donors was seventy-one years (range, thirty-four to ninety-four years), and there were four women and five men.
Grafts for the posterior cruciate ligament reconstruction were prepared from fresh-frozen whole patellar tendon allografts (Musculoskeletal Transplant Foundation, Edison, New Jersey). Bifid bone-patellar tendon-bone grafts were fashioned according to the technique described by one of us (J.K.S.) and colleagues3. The central 18 mm of the graft was isolated, and two bundles of 10 and 8 mm in width were fashioned to recreate the anterolateral and posteromedial bundles, respectively (Fig. 1). The patellar bone was used for the femoral attachments and was split along the line of the tendon fibers between the two bundles. Each patellar bone block was then sized to a length of approximately 2.5 cm and to fit a tunnel size of 10 or 8 mm, corresponding to the tendon width. Two 2.0-mm holes were then drilled into each femoral bone plug, and passing sutures of number-5 Ethibond (Ethicon, Somerville, New Jersey) were placed through each hole. The tibial bone plug was approximately 20 mm long, 13 mm wide, and 12 mm thick. A single 4.5-mm gliding hole was drilled into the center of the plug, through which a 4.5-mm fully threaded cortical screw (Synthes, West Chester, Pennsylvania) was inserted to fix the graft in the trough.
Surgical Technique
Double-Bundle Reconstruction
A posteromedial approach to the posterior part of the tibia was carried out through a horizontal incision in the popliteal crease. The osseous attachment landmarks of the posterior cruciate ligament were identified, a vertical arthrotomy was created between them, and the remaining posterior cruciate ligament insertion was identified and débrided. A narrow osteotome and burr were used to create a trough 13 mm wide and 10 mm deep that extended 2 cm distally from the posterior cruciate ligament insertion. The graft was secured in the trough with the 4.5-mm fully threaded cortical screw and washer.
The medial parapatellar arthrotomy permitted visualization for placement of the femoral tunnels. Beath pins were drilled from inside out for both the anterolateral and the posteromedial bundle. The footprint of the posterior cruciate ligament stump on the femur was used as a landmark for tunnel placement. In a right knee, the anterolateral bundle was placed at approximately the twelve-thirty position, approximately 5.5 mm from the articular margin, and the smaller posteromedial bundle was placed at approximately the three o'clock position, approximately 6 to 7 mm from the articular margin. Cannulated reamers were drilled over the Beath pins to create a 10-mm anterolateral tunnel and an 8-mm posteromedial tunnel. An anteromedial incision was made to dissect down to the pins in the medial femoral condyle. The anterolateral and posteromedial grafts were then passed through the femoral tunnels with use of a suture passer.
A previously detached bone block containing the posterolateral corner attachments (see below) was reduced, and a single 4.5-mm lag screw with a washer (Synthes) was used to reattach these structures prior to tensioning and fixation of the posterior cruciate ligament graft. Five to ten cycles of flexion and extension were performed to pre-tension the graft. Both bundles were fixed with metal interference screws (Smith and Nephew, Andover, Massachusetts) placed from outside in with the knee in 90° of flexion. A gentle anterior drawer to recreate the normal anteromedial step-off was maintained throughout femoral fixation.
Following testing of the fully reconstructed knee (double-bundle reconstruction with posterolateral corner repair), the screw securing the posterolateral corner osteotomy site was removed and the fragment was freed from its osseous bed. The specimens again underwent the same testing protocol (double-bundle reconstruction without posterolateral corner repair).
Single-Bundle Reconstruction
In order to test the single-bundle construct, the fixation was removed from the posteromedial bundle and all tension was released from this bundle by passing a probe antegrade through the posteromedial tunnel. This left only the 10-mm anterolateral bundle intact. The posterolateral corner was then replaced and fixed with a screw and washer (single-bundle reconstruction with posterolateral corner repair). Following repeat testing, the screw and washer were removed from the posterolateral corner and the specimens underwent the final evaluation (single-bundle reconstruction without posterolateral corner repair).
Testing Protocol
All knees were first tested while they were intact. Planned incisions, including the 2-cm medial parapatellar arthrotomy and the lateral approach to the knee, were performed in the intact specimen to avoid biasing the deficient states. The testing protocol consisted of posterior drawer measurements at 90° of knee flexion, stress radiography, and dial testing with a standard goniometer at both 90° and 30° of knee flexion. The interobserver and intraobserver reliability of the goniometer used for dial testing was evaluated at both 30° and 90°. A paired t test with significance set at p < 0.05 demonstrated no significant interobserver differences in the measurements made at 30° (mean and standard error, 13.6° ± 1.4° compared with 13.9° ± 2.6°; p > 0.05) or 90° (12.8° ± 1.9° compared with 12.6° ± 1.4°; p > 0.05). Analysis of variance also showed no significant intraobserver difference in the dial testing measurements at 30° (13.6° ± 1.4°; p > 0.05) or 90° (12.8° ± 1.9°; p > 0.05).
Posterior drawer measurements were graded as 0, 1, 2, or 3 on the basis of the amount of step-off at the medial joint line. Our grading system followed that of the International Knee Documentation Committee (IKDC), with Grade 0 indicating 0 to 2 mm of posterior translation; Grade 1, 3 to 5 mm; Grade 2, 6 to 10 mm; and Grade 3, >10 mm8. External rotational measurements were obtained with a standard goniometer and referenced from a previously placed drill-bit. Stress radiographs were made with the Telos stress radiographic device (Telos, Weiterstadt, Germany). The knees were placed in the device at a 90° flexion angle, and the device was then used to apply a posteriorly directed force of 200 N to the proximal part of the tibia. A metallic disk of known size was used to calculate radiographic magnification. Lateral stress images with the femoral condyles superimposed were then made with a fluoroscopic image intensification unit and printed for later measurement. Measurement of the stress radiographs was performed with use of the fixed landmarks described by Stäubli et al.9. The medial and lateral compartments were measured separately with a digital micrometer; the measurements were corrected for magnification and then averaged to calculate the average posterior displacement.
The posterior cruciate ligament, including the meniscofemoral ligaments when present, was then resected under direct visualization through the medial parapatellar arthrotomy. To simulate a posterolateral corner injury, an incision was made over the lateral femoral condyle. The iliotibial band was split longitudinally, and the origin of the lateral collateral ligament and insertion of the popliteus was identified. This permitted an osteotomy of the entire block of bone including both structures. Before the osteotomized fragment was elevated, a 4.5-mm hole was drilled in the center to permit precise reapproximation of the bone block to simulate reconstruction of the posterolateral corner without introducing variability in tensioning of the corner reconstruction.
Data Analysis
Both one-way repeated-measures and two-way mixed-design analysis of variance were used to determine differences in the dependent variables (external rotation at 30°, external rotation at 90°, and posterior displacement seen on stress radiography) between the knee conditions (intact, double-bundle reconstruction with posterolateral corner repair, double-bundle reconstruction without posterolateral corner repair, single-bundle reconstruction with posterolateral corner repair, and single-bundle reconstruction without posterolateral corner repair) with the significance level set at p < 0.05.
A prestudy sample-size calculation was performed to detect a clinically relevant difference between groups of 2.5 mm of posterior tibial translation with alpha set at p = 0.05. A sample size of ten specimens provided a power of 0.91, and a sample size of eight specimens provided a power of >0.80.
Posterior Drawer Testing
The posterior drawer measurements after the double-bundle reconstruction without posterolateral corner repair were Grade 0 in eight specimens and Grade 1 in one specimen. The single-bundle reconstruction without posterolateral corner repair resulted in a posterior drawer that was Grade 0 in one specimen, Grade 1 in seven specimens, and Grade 2 in one specimen. The addition of the posterolateral corner repair resulted in a Grade-0 posterior drawer in all specimens with a double-bundle reconstruction and in seven of the specimens with a single-bundle reconstruction. The addition of the posterolateral corner repair resulted in a Grade-1 posterior drawer in two specimens with a single-bundle reconstruction (Table I).
Dial Testing
Posterolateral Corner Intact
A comparison of the results of dial testing between the double and single-bundle reconstructions with the posterolateral corner structures intact is shown in Figure 2, and the means, standard errors, and 95% confidence intervals are shown in Table II. Compared with the intact knee (p = 0.01) and the single-bundle reconstruction (p = 0.027), the double-bundle reconstruction resulted in significantly decreased external rotation at 30° of knee flexion. However, there was no significant difference in external rotation between the single and double-bundle-reconstruction groups at 90° (p = 0.209).
Posterolateral Corner Deficient
A comparison of the results of dial testing between the double and single-bundle reconstructions without posterolateral corner repair is illustrated in Figure 3, and the means, standard errors, and 95% confidence intervals are shown in Table III. The double-bundle reconstruction permitted significantly less external rotation than did the single-bundle reconstruction at 30° (p = 0.03). However, without the addition of the posterolateral corner repair, the double-bundle reconstruction did not restore rotational stability to that in the intact controls at 30° (p = 0.014). At 90°, both the double and the single-bundle reconstructions without posterolateral corner repair allowed increased external rotation compared with that in the intact knee (p < 0.001), but there was no difference between the external rotation values associated with the two reconstruction techniques (p = 0.253).
Stress Radiography
The results of stress radiography are shown in Figure 4, and the means, standard errors, and 95% confidence intervals are shown in Table IV. Again, comparisons were first made with the posterolateral corner structures intact. The mean posterior displacement (and standard error) after the double-bundle reconstruction with posterolateral corner repair was 3.3 ± 1.4 mm. This was not significantly different from the mean posterior displacement of 4.8 ± 1.0 mm after the single-bundle reconstruction with posterolateral corner repair, and both values were similar to that in the intact controls (2.9 ± 0.5 mm) (p = 0.254). However, differences were noted between the values associated with the single and double-bundle reconstructions without posterolateral corner repair. The posterior displacement after the double-bundle reconstruction without posterolateral corner repair measured 6.4 ± 1.7 mm, and the corresponding value after the single-bundle reconstruction was 9.0 ± 1.9 mm. The value after the double-bundle reconstruction did not differ from the value in the intact controls (p = 0.318). However, the value after the single-bundle reconstruction was significantly increased compared with that in the intact controls (p = 0.039) and that after the double-bundle reconstruction (p = 0.026).
The aim of this study was to use familiar clinical tests to compare single and double-bundle tibial inlay reconstructions of the posterior cruciate ligament in a cadaver model. While evaluation of posterior displacement on stress radiographs and posterior drawer testing showed minimal differences between the two techniques, the double-bundle reconstruction provided more rotational stability than did the single-bundle reconstruction. This difference was significant on dial testing at 30° of knee flexion and could be predicted on the basis of the known function of the posteromedial bundle. While single-bundle reconstruction restores knee biomechanics at mid-to-high knee flexion angles, there still tends to be residual laxity when the knee is near full extension, according to some studies1,10,11. Recent biomechanical studies have shown that the addition of the posteromedial bundle reduces posterior tibial translation in knee flexion as well as in extension, suggesting that this bundle serves an important role throughout knee flexion1,2,12. We do not believe that this is solely due to the differences in the bulk of the respective grafts since, in this study, the stress radiography findings did not differ significantly between the two types of reconstruction. Rather, we believe that the better rotational stability results from a more anatomic restoration of the kinematics of the native posterior cruciate ligament.
The double-bundle reconstruction used in this study is a large graft intended to closely reproduce the normal posterior cruciate ligament anatomy3-5. Bergfeld et al. recently performed a biomechanical comparison of single and double-bundle reconstructions6. They utilized one-half of an Achilles tendon allograft for both techniques, in order to avoid biasing the study toward the double bundle as a result of the larger graft size. It has not yet been determined whether the benefit of a double-bundle reconstruction is weighted more heavily toward the orientation of the bundles or the bulk of the graft, but it is likely a combination of both. As was the case in the present study, Bergfeld et al. found no difference in posterior translation; however, they did not perform any rotational testing, and their model was an isolated posterior cruciate ligament deficiency model without injury to the posterolateral corner, which uncommonly requires surgical reconstruction.
One concern about using a double-bundle graft is that the creation of two tunnels in the medial femoral condyle possibly weakens it and predisposes it to fracture. We are not aware of any reports of this problem in the literature, nor have we encountered it in our clinical experience. On a purely mathematical basis, one would expect that two smaller-diameter tunnels would be better than one large-diameter tunnel since the volume of a cylinder depends on the square of the radius of the tunnel. Hence, a 20-mm-long tunnel of 18 mm diameter would require removal of almost twice the bone needed to create two 20-mm-long tunnels of 10 and 8 mm diameter. Nonetheless, we did not examine any failure characteristics related to the tunnel configuration in the femoral condyle.
We are somewhat concerned that the double-bundle reconstruction overconstrained the knee. Dial testing at 30° showed significantly less rotation after the double-bundle reconstruction than either in the intact control or after the single-bundle reconstruction. Interestingly, in a recent biomechanical comparison between single and double-bundle reconstructions in a cadaver model, Markolf et al. found that knees with a double-bundle reconstruction had a significant reduction in laxity at lower flexion angles (p < 0.05) but at the expense of higher forces in the posteromedial graft throughout the range of motion13. In spite of the increase in posterior stability of the knees with the double-bundle reconstruction, these authors thought that the increased forces in the posteromedial bundle may predispose the graft to elongation and questioned the rationale for double-bundle reconstruction. The increased forces in the posteromedial graft could certainly account for the overconstraint noted in our study. However, like Bergfeld et al.6, Markolf et al. did not perform rotational testing. The benefits of a double-bundle reconstruction may not be realized in the context of an "isolated" posterior cruciate ligament injury, for which nonoperative management is usually chosen, but they may become important in the context of subtle posterolateral corner injuries when reconstruction is indicated.
Our method of graft tensioning may have been a factor in the overconstraint that we noted after the double-bundle reconstruction. We chose to tension both grafts at 90°. Recently, Carson et al. evaluated different tensioning protocols and found that tensioning the posteromedial bundle at 90° and the anterolateral bundle at 0° reproduced anatomic in situ graft forces14. That study suggests that, in our study, there would have been increased forces in the posteromedial bundle, which might account for the observed overconstraint. However, the literature provides justification for tensioning the graft at knee flexion angles ranging from 0° to "high flexion" since the posteromedial bundle tightens both at extension and at high flexion1,15,16.
Notably, without repair of the posterolateral corner injury, the double-bundle reconstruction continued to resist posterior displacement in a manner comparable with the situation in the intact knee whereas the single-bundle reconstruction was significantly more lax. Rotational differences were again noted at 30° of knee flexion. One recent clinical study showed 90% good or excellent results at two years after isolated double-bundle reconstruction, even in the setting of Grade-1 or 2 posterolateral corner laxity17. The tendency for recurrent laxity of posterior cruciate ligament reconstructions is well known11,18. In one recent prospective study of the outcomes of single-bundle tibial inlay reconstruction for the treatment of isolated and combined injuries, Cooper and Stewart reported a Grade-0 posterior drawer in only nine of forty-one patients available for follow-up and an average side-to-side difference of 4.11 mm on Telos stress radiography18. Even though clinically detectable laxity did not contribute to patient dissatisfaction, the authors acknowledged that their results may be improved with use of a double-bundle reconstruction. In addition, the importance of treating posterolateral corner knee injuries along with posterior cruciate ligament injuries has been well described in the literature19-29.
Nevertheless, the failure to detect and treat a posterolateral corner injury did not appear to have the same effect on the biomechanical performance of a double-bundle reconstruction in our study. It follows that, in knees with combined injuries, an isolated single-bundle graft would perhaps be subjected to more stress and an increased risk of failure than would an isolated double-bundle graft reconstruction. Nonetheless, we strongly advocate the proper diagnosis and subsequent surgical management of all posterolateral corner injuries that are associated with posterior cruciate ligament injuries, regardless of whether a single or double-bundle posterior cruciate ligament reconstruction is performed.
There is a lack of comparative clinical studies of single and double-bundle posterior cruciate ligament reconstructions. In the only published, prospective, nonrandomized comparison of double-bundle and single-bundle reconstructions of which we are aware, Wang et al. did not detect a significant difference between the techniques at a minimum of two years postoperatively30. However, the double-bundle technique used by those authors is quite different from the one employed in the present study. They described using autogenous hamstring grafts for both the single and the double-bundle reconstructions. Their double-bundle reconstruction would, therefore, be much smaller than either the native posterior cruciate ligament or the double-bundle reconstruction described by one of us (J.K.S.) and colleagues3 as well as by others4,5, making comparisons difficult. Wang et al. specifically excluded posterolateral corner injuries from their group of high-energy posterior cruciate ligament injuries; thus, in light of our data, that may have minimized differences between the techniques. Clearly, there is a need for further clinical evaluation. Our study is unique in that we used common clinical tests in a cadaver model, thus providing a link between this in vitro and future in vivo studies.
Limitations of this study are those inherent in most biomechanical in vitro time-zero cadaver studies. The mean age of the donors of the specimens was seventy-one years. We were unable to control for bone quality, and that may have affected fixation of the grafts. However, both reconstructions were performed in the same randomly selected specimen from a matched pair, and thus we would expect that they would have been equally affected by the bone quality. Furthermore, both reconstructions were compared with the same intact knee as the control. The double bundle was tested first, which may have fatigued the anterolateral bundle, which we did not alter for the single-bundle tests. However, we do not believe that the force applied in any of these clinical tests was of a magnitude sufficient to cause clinically detectable fatigue or failure of the construct. Markolf et al. evaluated graft elongation with loads from 20 to 200 N and demonstrated a mean lengthening of a tibial inlay graft of 5.9 mm after 2000 cycles31.
We recognize that the double-bundle reconstruction permitted even less rotation than that observed in the intact knee at one flexion angle and that the single-bundle reconstruction allowed rotation similar to that in the controls when the posterolateral corner was intact. With graft tensioning, we aimed to achieve the most stable construct at time zero given the well-known tendency of grafts to elongate (particularly with multiligament reconstructions). In spite of the similarity in posterior translation, the double-bundle reconstruction did overconstrain the knee at 30° of flexion when the posterolateral corner was intact. It is unclear what effect this overconstraint might have if it persisted in the clinical setting, and the potential for accelerated osteoarthritis due to increased joint forces is a concern. However, one apparent important advantage of this degree of control was the ability of the double-bundle reconstruction to continue to resist posterior displacement in spite of the absence of the posterolateral corner structures. We believe that this increased time-zero stability may be beneficial in view of the common clinical finding that these reconstructions tend to stretch out over time in vivo.
We used familiar clinical tests, instead of sophisticated robotic devices, to evaluate our techniques. While the quality of data derived with mechanical testing devices is usually high, translation to the clinical setting is difficult. By using familiar clinical tests in this cadaver model, we hoped to begin to bridge the gap between in vitro and in vivo studies. We were able to define a set of parameters that may be more directly applicable to clinical practice and may therefore better direct clinical decision-making. The application of physical examination maneuvers in a nonclinical setting without muscle loads may appear to be a limitation of this model. However, in a clinical setting, the physical examination is often limited by patient discomfort and guarding. Surgeons tend to place greater importance on the results of examinations done with the patient under anesthesia, with complete muscle relaxation, and to base surgical decisions on those results. This is a situation similar to the model that we tested. Therefore, for this particular application of stress radiography and physical examination, the absence of physiologic muscle loads is an advantage of our injury model.
Lastly, a potential limitation of our study is our choice of methods to create and repair the posterolateral corner injury. Numerous techniques for reconstructing the posterolateral corner have been described in the literature26,32-38. In order to avoid confounding variables introduced by choosing a posterolateral corner reconstruction technique, we elected to elevate the femoral attachment of the lateral collateral ligament and popliteus tendon with an attached femoral bone block. This permitted simulation of a complete injury to the three important structures of the posterolateral corner—the lateral collateral ligament, the popliteus tendon, and the popliteofibular ligament (by virtue of its insertion on the now slackened popliteus tendon)—and it allowed a reproducibly anatomic repair by fixing the bone block back to where it had been detached.
The double-bundle open inlay technique for reconstruction of the posterior cruciate ligament provides more stability than does the single-bundle open inlay technique, primarily because of its contribution to increased rotational and posterior stability in the presence of an untreated posterolateral corner injury. Despite this superiority, a double-bundle reconstruction was not sufficient to completely restore knee stability without reconstruction of the corner, a finding that highlights the importance of addressing the posterolateral corner in combined injuries. Also, overconstraint at 30° of flexion after double-bundle reconstruction with repair of the posterolateral corner may be an important risk factor for osteoarthritis. Nonetheless, we believe that the increased time-zero stability afforded by the anatomic second bundle may be beneficial in view of the common clinical finding that these reconstructions tend to stretch out over time in vivo. 