Introduction

Articular cartilage defects appear frequently in all age groups and may cause severe limitations in activities of daily living due to symptoms such as pain and joint stiffness. With the introduction of autologous chondrocyte implantation (ACI) in 1994 [8] and the developments of this technique, it now appears that a promising treatment option may have been found. Many studies have been published, indicating that, if current indications and contraindications are respected [5], patients benefit significantly from this procedure even over the long term [7, 36, 40]. Furthermore, the second-generation ACI is also a suitable treatment option for the repair of focal degenerative defects of the knee [26].

To date, there has been no general agreement on the postoperative treatment and rehabilitation after ACI. There have been continuous surgical advances in the treatment of articular cartilage lesions, yet few detailed biomechanical evaluations after ACI exist. Previously published biomechanical studies analysing muscle strength after anterior cruciate ligament reconstruction have revealed significant strength deficits that persist at long-term follow-up [3, 4, 18, 23, 30]. Quadriceps weakness in particular, which may result from activation failure as well as muscle atrophy [39], has been shown to have a long-term negative impact on knee function and may also be a predictor of early onset of osteoarthritis [38, 44].

This is corroborated by Mikkelsen et al. [34] who showed that patients with good quadriceps torque after ACL reconstruction returned to their previous activity level significantly earlier.

To date, only five studies have been published dealing with muscle strength measurements after autologous chondrocyte implantation [13, 27–29, 31]. However, this is the first study to evaluate the particular influence of defect location on biomechanical strength outcomes and the balancing of different muscle groups.

The aim of this prospective study was to evaluate whether defect location had an impact on clinical outcome and isokinetic muscle strength performance 4 years after autologous chondrocyte implantation by comparing the operated leg with the healthy, non-treated extremity. Correlation between clinical and isokinetic outcome was done.

The results of this study may contribute to develop specific rehabilitation protocols optimized for different implantation locations to improve clinical and functional outcomes after autologous chondrocyte implantation.

Materials and methods

Between December 2001 and April 2004, 79 patients with focal posttraumatic or degenerative symptomatic cartilage defects were treated with matrix-associated autologous chondrocyte implantation (BioSeed®-C, BioTissue Technologies GmbH, Freiburg, Germany).

In total, there were 44 patients, who fulfilled the inclusion criteria (below) and were enrolled in this prospective trial for a period of 48 months.

Inclusion criteria

Patients had to be between 18 and 55 years of age. Chondral defects had to be full-thickness defects grade III B, C according to the ICRS classification system [19]. The lesions were sized ≥2 cm2 and had to be localized in only one of the 3 knee compartments.

Exclusion criteria

Patients with acute trauma, limits in knee extension or flexion less than 130°, varus or valgus deformities of more than 5°, patella malalignment with medial or lateral shift of more than 0.5 cm, instabilities of the collateral or cruciate ligaments, meniscal tears, status after total or partial meniscectomy, osteochondral defects, drug treatment for osteoarthritis or intra-articular corticosteroid injections within the previous month were excluded.

Table 1 shows the baseline characteristics of the included patients. The patient collective was divided into two major groups according to the defect location: the femoral condyle group (medial and lateral femoral condyles) (n = 28) and the patellofemoral joint group (Trochlea (n = 5) and retropatellar (n = 11)) (n = 16). There were no significant differences between both groups concerning patients’ characteristics.

Table 1 Patients’ characteristics: patients separated by the two main groups “femoral condyles” and “patellofemoral joint”
Full size table

Surgical procedure

Implantation of matrix-associated autologous chondrocytes was done in a two-step procedure: in a first arthroscopy, approximately 250 mg of healthy cartilage tissue was harvested from a less weight bearing area of the knee joint. After in vitro expansion, approximately 20 million chondrocytes were rearranged in a three-dimensional polymer-based scaffold (3 × 2 × 0.2 cm). The second step consisted in a mini-open implantation of the graft into the defect area by the previously described anchor-knot technique [15, 26, 27].

Rehabilitation protocol

Patients’ mobilization on crutches with a maximal loading of 15 % body weight and continuous passive motion and range of motion (ROM) exercises started the day after surgery. Patients with femoral condyle implants were not limited in their ROM in contrast to patients treated with patellofemoral joint implants. These patients were limited to 40° ROM for the first 2 weeks, 60° ROM in the following 2 weeks and 90° ROM in weeks 5–6 postoperatively. Patients then gradually increased loading and performed active physiotherapy and gentle ergometric training for the next 6 weeks. Full weight bearing was anticipated after 10–12 weeks in combination with increased strength training and coordination exercises. After 4 months, low-impact activities such as cycling or jogging were allowed, while more strenuous activities and contact sports were not recommended before 9–12 months after implantation. Patients’ compliance with rehabilitation was not monitored.

Clinical evaluation and isokinetic muscle strength measurements

Clinical outcome of 44 patients was evaluated using the International Cartilage Repair Society (ICRS) and the International Knee Documentation Committee (IKDC) forms [2]. Data were documented preoperatively and at 6, 12 and 48 months after implantation.

Strength measurements were performed after 48 months by a single examiner using an isokinetic dynamometer (CON-TREX MJ Multi-joint Module, CMV AG, Dübendorf, Switzerland). Isokinetic strength measurements performed on this dynamometer have been proven objective, reliable, valid and reproducible [9, 32].

On physical examination before isokinetic testing, none of our patients had significant effusion, swelling or pain on exertion, thus allowing us to perform strength measurements on all of them. Before testing, patients warmed up on a treadmill for 10 min. Patients were then seated in the assessment chair with trunk and thighs stabilized with belts to avoid additional body movement during isokinetic testing. The test protocol consisted of 5 reciprocal flexion and extension movements with maximal strength effort to evaluate hamstring and quadriceps strength. After gravity correction of the tested limb, each measurement was performed in three different test modes with the healthy, then the operated knee joint. Two concentric modes had an angular velocity of 60°/s and 180°/s and an eccentric mode 60°/s. Range of motion (ROM) was defined 0/10/90° (extension/flexion) at all test modes. A rest period of at least 60 s followed each set. In every mode, 5 maximum peak torque values were recorded in Newtonmeters (Nm).

To achieve the highest possible accuracy of our measurements, the parameter used for analysis of maximum strength was the mean value calculated from the best 3 of 5 maximum peak torque values as described in previous studies [21, 33, 35]. Furthermore, mean values were used to calculate the hamstring–quadriceps (H–Q) ratio of the operated leg in comparison with the healthy leg.

Statistical analysis

For statistical analysis, comparison of clinical and isokinetic outcomes between subgroups was performed using the Mann–Whitney U test for nonparametric data and considered significant at p < 0.05. Improvement or decline of clinical scores over the follow-up period was compared using the Wilcoxon test for parametric data. Differences were considered significant at p < 0.05. The Wilcoxon test was further used to analyse parametric biomechanical data at 4-year follow-up, performed on the healthy and operated knee of each patient. Differences were considered significant at p < 0.05. We also correlated clinical scores and isokinetic outcome at 48 months after implantation using the Pearson’s coefficient of correlation that was considered significant at p < 0.05.

Results

Clinical evaluation 4 years after implantation of autologous chondrocytes

The ICRS score (Table 2a) showed a constant improvement in both the femoral condyle and the patellofemoral joint group. Statistical comparison between both defect groups revealed no significant differences over the entire follow-up period.

Table 2 Clinical evaluation after 48-month follow-up (femoral condyles n = 28, patellofemoral joint n = 16)
Full size table

With a single exception (patellofemoral group 6–12 months), improvements were significant (p < 0.05) over the entire study period. Compared to the preoperative baseline, both groups achieved highly significant improvements (p < 0.001) 4 years after treatment. Mean findings in the ICRS score improved in both groups to now “nearly normal” clinical findings compared to “severe abnormal” before ACI (Table 2a).

The IKDC knee evaluation score (Table 2b) also improved continuously over the entire study period in both groups. Again no significant differences were found between both groups. The Wilcoxon test showed significant improvement for every tested adjacent follow-up period with highly significant improvement after 4 years compared to baseline results (p < 0.001).

Biomechanical strength evaluation 4 years after implantation of autologous chondrocytes

As Table 3 shows, the operated knee joint revealed significantly (p < 0.05) weaker quadriceps and hamstring peak torques compared to the corresponding healthy muscle group in all test modes and all groups. The patellofemoral joint group in particular showed severely reduced quadriceps strength compared to the femoral condyle group. However, even if the mean values of the patellofemoral joint group were noticeably worse, statistical comparison of single muscle groups did not show significant differences between both groups yet (p > 0.05). The hamstring–quadriceps ratio is often used in international literature to determine muscular balance. The physiological range is between 0.5 and 0.75 depending on age and angular velocity [1, 17, 20, 46], indicating that the quadriceps muscle normally is up to twice as strong as the hamstring group. Four years after implantation of autologous chondrocytes both groups revealed considerably altered hamstring–quadriceps ratios particularly on the operated extremity (Table 4). Furthermore, resulting from the aforementioned severe reduction in quadriceps strength, the operated leg of patients treated at the patellofemoral compartment experienced a change in strength with ratios above 1.0 in concentric and eccentric test modes. Statistical comparison of hamstring–quadriceps ratios now revealed significantly worse results in the patellofemoral joint group compared to the femoral condyle group (p < 0.05, concentric 60°/s operated and healthy, eccentric operated).

Table 3 Isokinetic mean values (femoral condyles n = 28, patellofemoral joint n = 16)
Full size table
Table 4 Hamstring–Quadriceps (H–Q) ratio (femoral condyles n = 28, patellofemoral joint n = 16)
Full size table

Correlation of clinical and isokinetic scores 48 months after implantation of Bioseed-C

Correlation between clinical scores and isokinetic values were performed for all patients together using the Pearson’s coefficient of correlation (PC). Objective ICRS score showed positive correlation (p < 0.05) with all collected peak torque mean values with a single exception (Table 5). The subjective IKDC score showed positive correlations (p < 0.05) with the operated extensor and flexor muscles in the concentric test modes.

Table 5 Correlation between clinical scores and isokinetic mean values (all patients, n = 44)
Full size table

Discussion

The most important finding of the present study was the persisting significant strength deficit at the operated extremity even 48 months after ACI and furthermore the location-specific significant differences in hamstring–quadriceps ratios.

Supporting previously published positive mid- to long-term results after the second-generation ACI [6, 12, 24], the included patients showed good to excellent clinical outcomes with highly significant improvement 4 years postoperatively compared to the preoperative baseline. Furthermore, the patellofemoral joint group, for which worse clinical outcomes had been previously reported [7, 37, 40], showed comparable results to the femoral condyle group with no statistically significant differences. However, despite the good clinical outcome, the patellofemoral joint group in particular showed the greatest strength deficits, predominantly affecting the quadriceps muscle.

The biomechanical data presented in this study confirm the results of two other studies including strength measurements after autologous chondrocyte implantation published by Loken et al. [31] and Ebert et al. [13]. In Lokens et al. study, 21 patients were evaluated clinically and isokinetically 1, 2 and 7.4 years after the first-generation ACI. While clinical scores improved constantly, they found highly significant side-to-side differences for knee flexion and extension workload even 7.4 years after ACI [31]. Supporting the findings of the present study, strength deficits of the operated knee in their patient population were greater in knee extension than in flexion. The authors did not distinguish between different defect locations.

In Ebert’s et al. [13] published study, isokinetic strength values of 60 patients 5 years after the second-generation ACI were assessed. While there were no significant differences in hamstring strength between operated and non-operated side, they found a significantly impaired quadriceps strength at a defined flexion angle of 45° on the operated leg.

While the statistical comparison of hamstring and quadriceps muscles individually did not show significant differences between both defect groups, the particular impact of the defect location could be revealed by analysing the muscle balancing. Resulting from the severely impaired extensor muscles, the patellofemoral group showed significantly worse hamstring–quadriceps ratios compared to condylar defects in all test modes (p < 0.05).

Ebert et al. [13] also reported elevated H–Q ratios varying from 0.81 to 0.93 on the healthy extremity and from 0.86 to 0.97 on the operated side. Bilaterally elevated H–Q ratios are suggestive of a general bilateral postoperative weakness in quadriceps strength that result from the prolonged period of relative inactivity after ACI surgery [13]. Ebert et al. did not provide information about their patients’ clinical outcome, and they only treated patients with femoral and not with patellar defects, what aggravates a comparison with our results.

A possible explanation for the significantly reduced quadriceps strength of the patellofemoral group might be the more invasive arthrotomy with elevation of the patella compared to defects on the femoral condyles. For the articulation of the patella, muscle balancing and proprioception in the patellofemoral joint are more important than in the femorotibial compartment. With the more invasive approach in the patellofemoral compartment, both the extensor muscles, and the joint capsule and ligaments were affected, leading to a pronounced strength deficit of the extensor muscles. Development of more soft-tissue sparing surgical approaches in this specific compartment might therefore reduce damage to these structures and may result in improved postoperative muscle strength and balancing.

Likewise, the delayed postoperative exercise regimen of this group (weeks 1–2 flexion limit 40°, weeks 3–4 flexion limit 60°, weeks 5–6 flexion limit 90°) may as well result in more muscle atrophy and functional limitation of the knee joint, than generally anticipated.

Strengths of our study are the detailed bilateral biomechanical evaluations 4 years after ACI and well-defined inclusion and exclusion criteria. In this context, patients with pre-existing patellofemoral or femorotibial malalignment, which have been shown to influence clinical and functional outcome [45], were previously excluded, thus raising the quality of the presented results.

A limitation of the present study is a follow-up intermission between the results after 1 year and the midterm re-evaluation 48 months after ACI. However, based on previous studies after ACI, which showed greatest improvement in clinical outcome predominantly in the first 12 months and only few changes afterwards [12, 16, 42], we decided to re-evaluate the patient collective with special focus on midterm results at 48 months.

Furthermore, we did not differentiate if the operated knee joint was in the dominant or non-dominant leg, based on previously published studies, that showed no significant differences between dominant and non-dominant leg [11, 13].

Strength measurements were not compared with a non-operated control group. As an alternative, the patients’ contralateral extremity was used as a control, assuming that the non-treated side would correspond to a healthy normal value for each patient [13, 22, 31]. It cannot be conclusively determined if patients with retropatellar or trochlear defects had greater pre-existing strength deficits before implantation of autologous chondrocytes, since biomechanical tests on this isokinetic dynamometer were unfortunately not available earlier than 48 months after ACI. However, preoperative assessment of maximum strength scores tend to be unreliable because of pain, swelling and other functional limitations [13]. For this reason, both other available isokinetic studies after ACI by Ebert et al. and Loken et al. [13, 31] purposely abstained from performing preoperative muscle strength measurements similar to the present study, where the focus of the biomechanical tests was placed on the 48-month midterm outcomes.

Besides the good clinical midterm results of the included patients, the distinct impairment of quadriceps strength in the patellofemoral compartment with resultant muscular imbalance in the H–Q ratio is concerning. As reported in other studies, quadriceps strength deficit is a strong negative predictor for long-term clinical outcome, since it leads to early onset of osteoarthritis [41, 43]. For this reason, all efforts must be made to restore muscle strength, particularly of the quadriceps to secure normal muscle balancing. Additionally, the rehabilitation process should be appropriately modified to include early patella mobilization as well as postoperative strength and neuromuscular coordination exercises.

Affirming this aspect, Della Villa et al. [10] reported that early postoperative quadriceps training after ACI had positive effects on clinical outcome and resulted in an earlier return to preoperative activity levels. Furthermore, intensified physical training after ACI led to significantly better long-term outcomes compared to patients with a lower activity level [25].

The preoperative quadriceps strength is also an important factor for postoperative outcome: Eitzen et al. [14] showed in their study that patients with preoperative quadriceps strength approaching normal values had significantly better results 2 years after anterior cruciate ligament reconstruction. Therefore, we recommend quadriceps training in the preoperative period, e.g. between arthroscopical cartilage biopsy and implantation of the loaded scaffold 6 weeks later. The location-specific biomechanical strength measurements after ACI in the present study help to improve treatment and postoperative rehabilitation protocols of patients with patellofemoral defects. However, further studies examining the influence of such modified pre- and postoperative rehabilitation protocols on clinical and biomechanical outcomes after ACI are recommended.

Conclusion

The promising clinical results of our patients are negatively affected by a marked persisting strength deficit especially affecting the quadriceps muscle of patients treated in the patellofemoral compartment. Therefore, muscle strength must be restored as soon as possible in order to prevent clinical worsening. A pain adapted bilateral preoperative strength training and a defect location-specific postoperative treatment regimen under physiotherapeutic guidance with close clinical observations for the first two postoperative years would be desirable.