Effect of graft positioning on dissipated energy in knee osteochondral autologous transplantation—A biomechanical study

Focal cartilage defects can be treated by osteochondral autologous transplantation (OAT). High congruence of the graft with the surrounding cartilage structure is essential for a good clinical outcome, but can not always be achieved. We recently established a method to measure dissipated energy (DE) as a friction parameter in knee joints. We now investigated how autograft harvesting and implant positioning affect the DE during knee motion. Six sheep knee joints were cyclically motioned under 400 N axial load. During the cyclic motion, the flexion angle and the respective torque were recorded and the DE was calculated. Several experimental conditions were tested: first, the DE was measured after approach had been performed (“native”). Subsequently, a cylinder was removed from the medial femur condyles and a donor cylinder was inserted from an unloaded site in four different transplant positions: even, 1 mm deeper, 1 mm higher, and flush without cartilage (defect). No significant changes in friction were observed between the native knee and an even or deep OAT positioning. We, however, found a small but significant increase in DE between the “native” and “1 mm high” formations (ΔDE compared with native = 14 mJ/cycle; P = .004 after data normalization) and a large increase in defect situation (ΔDE compared with native = 119 mJ/cycle; P = .001). Considering the long‐term therapeutic aim that is pursued when performing OAT, elevated graft positioning should clearly be avoided. From a biomechanical point of view, donor site morbidity after cylinder harvest can be neglected.

the medial femoral condyle (32%), followed by the patella (22%). 3 Chondrocytes have a very low metabolic activity and lose their mitotic ability early in life. 4 Thus, the cartilage has only a very limited capacity for self-repair. 5 In case of traumatic cartilage injury surgical therapy is therefore often necessary.
Especially in small defects with possibly also subchondral pathology, osteochondral autologous transplantation (OAT) is the therapeutic option of choice. In contrast to bone marrow stimulating techniques and autologous chondrocyte transplantation it offers the advantage of a mature full-thickness cartilage transplant which is press-fit into the bone, 6 allowing early mobilization and a high rate of return to sports 7 and a superior level of athletic activity. 8 Compared with bone marrow stimulating techniques like micro fracturing the therapeutic effect is of a longer nature thus making it more interesting especially for younger and active patients.
In OAT, a donor cylinder is harvested from a non-loaded hinge region and inserted snug into a female cylinder of similar size created at the cartilage defect site in the loading zone. An important requirement for a good clinical outcome is a high congruence of the graft with the surrounding cartilage structure, which can not always be achieved. Recent clinical studies indicated that an elevated graft of more than 1 mm is poorly tolerated. 6 Biomechanical studies in this area have so far been either limited to pressure measurements 9,10 or to evaluation of the coefficient of friction, measured with a simple cable model dissecting all muscles and soft tissue. 11 We have recently established and validated a method to measure dissipated energy (DE) as a friction parameter in knee joints with local cartilage defects.
The key advantage of this model is that the measurements can be performed in complete joints thus ensuring experimental conditions resembling closely to the in-vivo situation without the need to extensively dissect the soft tissues. 12 The aim of the present study was to extend our model for  9 we hypothesized that especially the upstanding graft would lead to increased friction in the joint.

| Specimens
Six fresh-frozen sheep knee joints were obtained post mortem and directly stored at −20°C. Prior to testing the joints were thawed overnight at room temperature wrapped in a cloth soaked with physiological saline solution. During preparation and testing, knee joints were additionally sprayed with saline solution from outside to prevent the soft tissue from dehydration.

| Specimen preparation
Osteotomies of the femur and tibia were performed 20 cm above/ below the joint space. At a distance of about 8 cm distal/proximal to the osteotomy, the bones were completely cleaned of periosteum, embedded in two-component resin (RenCast© FC 53 isocyanate/FC 53 polyol, Gößl & Pfaff GmbH, Karlskron, Germany) and fixed within aluminum cylinders in the robot. The settings of the robot, axes definition and recording of the passive path remained unchanged as described in previous studies. 12,14 Directly before the measurements, we performed a medial parapatellar approach to the knee joint and opened the capsule. The remaining synovial fluid was removed by rinsing and drying and instead NaCl 0.9% was distributed in the whole joint. Rinsing was repeated after each measurement to obtain uniform lubrication ratios throughout the whole measurement procedure and to eliminate dehydration effects as a possible confounder.

| Surgical treatment
The medial parapatellar approach was then closed by suture to prevent dehydration during the measurements and to obtain a soft tissue situation similar as after surgery. In addition, during each measurement, the knee joint was sprayed with NaCl solution and loosely wrapped with thin transparent film.
In the next step, recording of the individual passive flexion path was performed, 15,16 which was required for further measurements.  Figure 1).
The osteochondral graft was inserted into the defect cylinder in four different conditions: first "even" (0 mm height difference), then "deep" (1 mm below the surrounding level), and finally "high" (1 mm above the surrounding level). These three conditions were randomized within the six specimens.
Simulating the wear of cartilage in an elevated graft position, we finally removed the cartilage of the cylinder implanted in the "high" condition with a surgical fraise ("defect"), so that the exposed bone and the surrounding cartilage were even (see Figure 2D).
To realize different graft positions, metal discs with a height of 0.5 mm were used at the bottom of the cylinder.
To prevent sinking in of the graft, especially in the "high" and "defect" condition, a screw was inserted from the medial femur to support the base of the osteochondral graft. The head of the screw was sunk medially into the cortical bone to exclude a soft tissue irritation of the DE measurement. In order to subsequently remove the metal discs and the osteochondral graft, an additional, thinner hole from the base F I G U R E 1 Fixed knee joint in the robot (blue arrow) after a medial parapatellar approach. Extraction of the graft with the receiver trephine (red arrow) from an unloaded zone [Color figure can be viewed at wileyonlinelibrary.com] F I G U R E 2 A, flush graft position without a difference to the surrounding cartilage ("even"), B, recessed graft placement ("deep"), C, 1 mm elevated graft positon with medial fixation screw ("high"), and D, flush graft position after cartilage resection ("defect") [Color figure can be viewed at wileyonlinelibrary.com] of the transplantation site to the medial cortical bone was placed.

| Data analysis
The knee movement varies for ±10°about a flexion angle of approximately 60°. The axis of rotation varies with the vertical load, the defect grade, and the flexion angle. The intersegmental force and moment are functionally meaningful if they are defined at the "joint center" that lies on the axis of rotation. 17 Therefore, the screw axis identification method was used to determine the instantaneous screw axis parameter for each displacement Therefore we do not consider interfering factors, such as micro fragments or dehydration caused by the procedure to be of relevance. Harvesting the donor cylinder thus also does not appear to influence the DE in the knee joint.
Second, we investigated the effect of different types of graft positioning using DE as experimental outcome parameter. We thus analyzed and compared the DE for the conditions "native," "even," "deep," "high," and "defect" ( where significant changes could be detected between the five conditions (native, even, high, deep, defect) (Friedman test χ 2 (4) = 20.400; P < .001; n = 6).
We then performed Dunn's pairwise post-hoc tests and Bonferroni adjustments of the P-values on the basis of an alpha of .05. Significant differences were found between "native" and "defect" (P = .001; ΔDE = 119.38 mJ/cycle), as well as between "even" and "defect" (P = .010; ΔDE = 112.01 mJ/cycle). The effect sizes (r) were calculated using the formula r = z/√n (z = test statistic; n = number of pairs). A strong effect (r > 0.50) was found for "native" vs "defect" (r = 1.06) and "even" vs "defect" (r = .95).
To address the wide variation of values between the different knees in the "native" condition, the data were adjusted to their Median of the native condition. To this end, an adjustment factor k N was calculated for each knee using the formula: k N = 100-m DE (m DE is median of 18 cycles in the condition "native"). For further calculations, the factor k N was added to the median of each condition and each knee ( Table 2).
When comparing the levels of DE between the three positioning conditions (even, deep, high) based on the adjusted data, a significant difference was observed across the groups (Friedman test χ 2 (2) = 10.33; P = .006; n = 6) (see Figure 5). Again post-hoc analysis was performed with Dunn's pairwise tests and Bonferroni adjustment was performed based on an alpha of .05. A significant difference was then found between the conditions "even" and "high" (P = .004; ΔDE = 14.09 mJ/cycle). When calculating the effect size (r) as described above, a strong effect (r > .50) was found for "even" vs "high" (r = 1.06).

| DISCUSSION
Our results indicate an enormous rise in friction in the flush graft position in the defect situation, where values for DE were almost three times higher (+165% compared with "native"). Likewise, the increase of DE in the condition "high" was still 22% when compared to "native." Only a small increase of 9% was found in the "deep" condition compared with "native." In recent decades, pressure distribution in the joint had been the gold standard to evaluate the effects of graft positioning.  Over the past few years, in addition to pressure measurements, friction has also been established as an outcome parameter for OAT. Another interesting result of our study is the similarity of the DE in the conditions "native" and "even." A main criticism of OAT is the high donor-site morbidity which occurs clinically in the form of patellofemoral disturbances, crepitation, knee stiffness, and persistent pain. 20 Lane et al. describe an increased friction by a factor of about 1.5 to 2 even in an "even" graft position resulting from the surgical procedure including donor-site morbidity. 11 In our own study, we only observed an increase of DE from nativ to even of 8% without any statistical significance. We also attribute this minor difference in comparison to the results from Lane to the more physiological load protocol we applied. Our results may be an indication that the clinically reported symptoms are more likely caused by soft tissue irritation as a result of the surgical approach and scar formation rather

| Study limitations
The measurements in the in-vitro model are limited to the moment after surgery. In the long run, arthroscopic examinations showed that in the "deep" condition, fibrocartilage-like tissue can fill the depressed area and lead to an optically smooth surface. 13 While DE data on these fibrocartilage-like situations are still warranted, an older biomechanical study analyzing dog knees 11 months after creating a 6 mm chondral defect found no significant differences in mean and peak pressures between fresh and healed defects. 26 This could be an indication that fibrocartilage healing does not lead to a pronounced change in biomechanical conditions.
It also needs to be pointed out that while the "high" condition can especially, in the long run, lead to cartilage destruction as mentioned above, the effects of friction for these biological changes remain unclear.

| CONCLUSION
With our experimental setup that closely imitated the physiological biomechanical conditions of the knee joint during the gait cycle we found no significant changes in friction between the native knee and an even or deep OAT positioning. From a biomechanical point of view, donor site morbidity after cylinder harvest can be neglected.
Since the defect OAT situation led to a tremendous increase in DE, considering the long-term therapeutic aim that is pursued when performing OAT, elevated graft positioning should be clearly avoided.

ACKNOWLEDGMENT
The study was funded by the department's internal resources without external sponsorship.

AUTHOR CONTRIBUTIONS
CW conceived the study, supervised the experiments and wrote the manuscript; DT performed the surgical procedures and experiments, AB and CJ wrote the robot application and helped with the statistical analyses; UKH helped with the statistical analysis and co-wrote the manuscript. All authors have read and approved the final submitted manuscript.