This study, to our knowledge the largest consecutive long-term series on the surgical treatment of traumatic knee dislocations, has produced the following major results.
Firstly, on average 12 years after one-stage open complete reconstruction of traumatic knee dislocations 3/5 of patients showed good to excellent subjective and objective results. On average only 10° loss of flexion was observed. 4/5 of patients returned to their previous work but, on average, the activity level reflected in the Tegner score relevantly decreased from injury to last follow-up. The IKDC score was normal or nearly normal in 3/5 of patients and the Lysholm and Knee Society score showed good to excellent results. The instrumented anterior and posterior laxity measurements with the KT-1000 arthrometer demonstrated good to excellent results at mean 12 years after injury. Less than 5% of patients in the present study had an abnormal (>6 mm) anterior or posterior laxity measured with the rolimeter. These results are consistent with the findings of Tzurbakis et al.[31] who found 1.6 ± 1.9 mm and 2.3 ± 1.7 mm of anterior and posterior translation side-to-side difference. Varus und valgus laxity did not appear to be a major problem. It was normal (<3 mm) in 86% of our patients. This finding is consistent with those obtained by others[19]. Clearly, the measured values of anterior-posterior and varus-valgus laxity reflect a stable knee status in the majority of our patients, which was one major goal of our reconstructive treatment protocol.
Comparison of results with other studies on the treatment of knee dislocations is difficult as study populations often differ with regards to age, injury pattern, demographics, body mass index and treatment protocol. Most authors agree that non-surgical treatment with cast immobilization produces inferior results compared to surgical treatment regimens [5, 17, 29, 43]. A variety of surgical procedures both open[8, 21, 27, 29, 31, 44] and arthroscopy-assisted[19, 21] have been reported.
Using a similar open treatment strategy as ours Tzurbakis et al. reported comparable functional results in terms of instrumented anterior/posterior laxity and Lysholm and Tegner scoring for a series of 25 patients with a follow-up of 2-8 years[31]. Richter et al. reported on a series of patients and compared the outcomes of surgically and non-surgically treated patients. They found inferior results in terms of the IKDC and Lysholm and Tegner scores, which may be explained by the high number of polytraumatized patients in their series[17]. As a referral center for severe knee ligament injuries we often see trauma patients after primary stabilization and care in another hospital, which partly explains the low rate of neurovascular injuries in our series.
With an arthroscopy-assisted combined ACL/PCL reconstruction technique Fanelli et al. reported excellent, mildly superior results with regard to the Lysholm score and Tegner activity score, but the follow-up was rather short[19]. Also Harner et al. and Owens et al. found comparable results for the Lysholm score, loss of flexion, and KT 1000 laxity of the ACL and PCL. However, less patients achieved normal or nearly normal on the total IKDC score[20]. Although in recent years there has been a shift toward arthroscopy-assisted techniques, to our knowledge, all these studies have recorded only equivalent or inferior results in terms of subjective and objective outcomes compared to our long-term series. In addition, several injuries such as posterolateral or posteromedial corner injuries, fractures or avulsed tendons cannot or should not be treated arthroscopically. Therefore, we still propose our treatment protocol (including arthrotomy and open surgery) in acute cases of patients with multiple ligament injuries as a valuable treatment option. In our view, the question of which surgical approach the orthopedic surgeon should choose, i.e. open or arthroscopy-assisted, is only of marginal importance. It is far more the experience and teamwork of the surgeons, physiotherapists and nurses involved in the treatment that makes the difference.
Secondly, we believe that in this study stress radiographs in 30° and 90° flexion obtained with the Telos device have, for the first time, shown convincing long-term laxity results for anterior and posterior translation in surgically treated patients after traumatic knee dislocation. To our knowledge, only Fanelli et al. have to date reported comparable findings with a mean side-to-side difference in posterior translation of 3.2 mm in posterior stress radiographs[19]. Our instrumented radiographic stress testing in both anterior and posterior directions confirmed the clinical impression of a stable ACL and PCL in our long-term series of patients.
Controversy persists as to whether early surgical repair of the ligaments decreases the incidence of posttraumatic osteoarthritis compared with non-surgical treatment. In the present study only a mild degree of osteoarthritis was evident in most of the patients treated according to the Kellgren Lawrence Osteoarthritis score. Only 11 patients (16%) showed a Kellgren Lawrence score of grade III or IV, namely moderate or severe osteoarthritis. This was also reflected by our joint space measurements. This is in accordance with Richter et al.[17] who reported mostly mild and moderate osteoarthritic changes after a similar follow-up time. Interestingly, more degenerative changes were found in the non-surgically treated group of patients, which may be attributed to abnormal joint kinematics due to ligamentous instability. In contrast, Almekinders et al. did not find any difference in radiologically visible degenerative changes of the knee joint between surgically and non-surgically treated patients[45].
Thirdly, we had to accept an unavoidable rate of persistent problems, secondary surgeries and reoperations in this heavily injured patient population. Unfortunately, several authors failed to report their rate of secondary surgeries and/or reoperations[21, 22, 24]. The biggest problem we encountered after reconstructive surgery was continued pain and loss of motion. A pain level VAS >3 was found in 9% of patients. Decreased ability to flex the knee <110° was present in 4 patients. Seven patients (10%) underwent an arthroscopic arthrolysis during follow-up, which is consistent with the literature[20, 27, 31].
Instability was not really an issue for most of the patients followed-up in our series, which is reflected by the fact that only 7 patients (10%) underwent a secondary ligament reconstruction during the follow-up period. Three of them were a consequence of insufficient stability after the previous surgery. All but one reoperated patient received ACL suturing and/or PCL refixation. The number of patients requiring high tibial osteotomy or with a primary unconstrained total knee arthroplasty was low considering the long-term follow-up.
Fourthly, according to our results injury pattern, timing of surgery and the chosen surgical treatment might play a crucial role in the long-term prognosis.
In the present study, long-term outcome was significantly influenced by the type of injury.
We found that patients with an injury of the LCL and/or the peroneal nerve had a higher need for workers compensation (p < 0.01). Oswald et al.[25] found similar results reporting that patients with an injury on the lateral side had a less favourable outcome than patients with an injury on the medial side.
The timing of surgery significantly influenced patient's need for workers compensation. Generally it depends on the vascular status of the extremity, soft tissue conditions, concomitant injuries, comorbidities and the injury pattern. There is no doubt that an irreducible knee dislocation, dislocations associated with popliteal artery injury, or open dislocations demand emergency surgical management. In all other cases, there is a general consensus to wait with reconstruction until the inflammatory response has subsided and the full range of motion has been regained. The risk of arthrofibrosis, a major adverse event associated with premature reconstruction, is considered to be less if surgery is postponed until 2-3 weeks after injury[28]. Owens et al. advocated a surgical reconstruction within 2 weeks of the injury and Rihn et al. within 3 weeks of the injury[7, 26]. Postponement of surgery beyond 3-4 weeks is not recommended as this may result in excessive scarring of the collateral ligaments[7, 28].
The answers to the question of whether a ligament should be reconstructed or not vary greatly in the literature. Most authors advocate a one-stage surgical reconstruction of at least both cruciate ligaments[8, 19, 20]. Only a few authors reported the results of a multistage procedure. In 2002, Ohkoshi et al. reported good range of motion and antero-posterior laxity (KT-1000 manual maximum) after a two-stage surgical approach (first stage 2 weeks after injury: reconstruction of posterior cruciate ligament, second stage three months later: reconstruction of ligaments that have not healed as a result of non-surgical treatment). However, only eight patients were included in this series[24].
The choice of surgical treatment was significantly associated with better or worse outcome. We found that patients who underwent ACL suturing had a less favourable outcome than patients with reconstruction. Our findings are comparable to those obtained by Mariani et al., who found that patients treated by direct repair had less favorable results in terms of laxity and range of motion than reconstructed patients [22]. Wong et al. found that surgical treatment involving the complete repair of all injured structures was superior in terms of IKDC score and antero-posterior laxity when compared to partial repair (unicruciate ligament reconstruction)[33].
The selection of grafts is still based more on surgeon's preference and availability of grafts than evidence[4]. Patellar tendon, quadriceps tendon and hamstring tendons of the ipsi- or contralateral side have been used as autografts for reconstruction in traumatic knee dislocations. Several authors prefer allografts (Achilles tendon, patellar tendon, tibialis anterior tendon, hamstring tendons) as surgical time and donor site morbidity may be decreased in this complex reconstructive setting[19, 23, 32]. To our knowledge there is no study demonstrating the superiority of allografts in terms of clinical outcome and ligament laxity in comparison to autografts in multi-ligament injured knees. In addition, allografts are hardly available in Europe. Synthetic grafts have occasionally been used in reconstructive surgery in patients with traumatic knee dislocations[24, 30]. A series of 20 patients was evaluated retrospectively by Talbot et al. with a minimum follow-up of one year, yielding inferior results in terms of the Lysholm score, range of motion and ligament laxity[30].
We are aware that this investigation is subject to all the problems inherent in a retrospective study setting, but investigating patients treated for traumatic knee dislocations is hardly feasible in a prospective way. As there is no control group we compared our results to the available literature. However, our study has an extraordinary follow-up rate and is a consecutive series of patients treated at our hospital over a period of 27 years. In the present study performing multiple univariate analysis of about 200 correlations might have led to false positive results in about 10 (p < 0.05) or 20 (p < 0.1) cases and hence the results should be interpreted with all due caution.