IKD patients were included between June 1, 2016, and May 31, 2020, in our department. The inclusion criteria were as follows: 1) age: 18–60 years old, BMI≤31; 2) KD-IIIM (multiligament knee injury with ACL, PCL, and MCL ruptured) (Solarino et al., 2015; Jr, 1994) (Figures 1A, B); 3) irreducible (the closed reduction could not be performed); 4) the two-staged surgery was performed; 5) the stage-1 surgery was performed in 1 week post injury. The exclusion criteria were as follows: 1) KD-IIIL (multiligament knee injury with ACL, PCL, and LCL + PLC rupture) (Jr, 1994; Solarino et al., 2015); 2) history of lower extremity fracture, ligament rupture, and operations; 3) rheumatoid arthritis, ankylosing spondylitis, Parkinson’s disease, or other nervous system diseases.

The protocols and procedures for the protection of human subjects were approved by the Ethics Committee in our hospital (IRB ethical approval: KY-2016-OB08), and all of the methods were conducted in accordance with the approved guidelines.

In the beginning, vascular and nerve injury assessment was performed on the patients. The two-staged operation was performed by the same senior surgeon. Knee X-ray and CT were performed before the operation.

The stage-1 operation consisted of urgent open reduction and extra-articular structure (MCL, medial retinaculum, and capsule) repair. With a pneumatic tourniquet inflated on the thigh, a 5-cm longitudinal incision was made proximally from the medial femoral epicondyle, distally to the lower border of pes anserinus. This reverted the skin and subcutaneous fat flap posteriorly and exposed the damaged medial structures (Figure 2A), which were incarcerated into the condylar notch (Figure 2B). The torn MCL, medial retinaculum, and capsule were replaced and meticulously sutured with interrupted #1 absorbable braided sutures. The ruptured MCL and medial gastrocnemius muscle from the bony attachment were reinserted at its origin with a suture anchor (Figure 2C). Finally, the dislocated joint restored normal alignment after stage-1 (Figure 2D).

The intermediate rehabilitation lasted for 6–8 weeks (Ng et al., 2020). An above-the-knee plaster was applied with the joint flexed to about 20°. The patients were asked to follow the instructions for non–weight bearing for 2 weeks. Then, partly weight bearing was allowed and a rehabilitation program was started to restore the joint range of motion (ROM) of 0–120° and muscle strength (3rd–6th week). Full weight bearing was allowed in the 6th–8th week. The MRI, clinical assessment, and self-administered questionnaire (Lysholm and International Knee Documentation Committee (IKDC) score (Ranger et al., 2011)) were assessed after the rehabilitation.

The stage-2 operation was begun when the rehabilitation was accomplished with an ROM of 0–120°. All patients underwent the arthroscopic reconstruction of ACL and PCL (Figure 2H). Unilateral semitendinous and gracilis tendons were used as auto-graft. The femoral side was fixed using EndoButton CL (Smith and Nephew, Memphis, TN, United States), and the tibial side was fixed using the poly-l-lactic acid biodegradable interference screw.

The follow-up was started when stage-2 surgery was completed. The end was re-dislocation/death/missing, whichever occurred first. The minimum follow-up was 0.5 years, and the maximum was 1 year. General parameters included age, gender, body mass index (BMI), occupation, trauma level, MCL injury type, time between injury and stage-1, time between stage-1 and stage-2, follow-up time, and complications. Knee X-ray photographs, clinical assessments, and self-administered questionnaires (Lysholm and International Knee Documentation Committee (IKDC) score (Ranger et al., 2011)) were performed at 0.5 years and 1 year follow-up. All of the subject data were checked and entered into the database by two researchers, and double entry was used to carry out quality control.

The physical examination for assessing joint stability of IKD included the Lachman test, posterior drawer test, pivot shift, and varus and valgus stress test (Ranger et al., 2011).

Ligament laxity was measured using KT-1000: the forward shift is tested when the knee is flexed at 30°, and the backward shift is tested when the knee is flexed at 70°. According to IKDC standards, laxity is evaluated by comparing to the healthy side: a difference of less than 3 mm is normal, 3–5 mm is close to normal, 6–10 mm is abnormal, and more than 10 mm is severe.

The ROM was measured using a standardized goniometry technique. The knee flexion contracture (KFC) was assessed by passive physical examination. KFC is defined as the gap value of extension loss compared to the normal side, and more than 5° was considered KFC according to the Knee Society Score (KSS) system (Insall et al., 1989).

The sample size was pre-calculated by using the following formula: n= (μ _α+ μ _β)² σ²/δ², a was set at 0.05, β was set at 0.01, and the power was 90% (1-β). The comparisons of the count data were processed by the chi-square test or Fisher’s exact test, and continuous data were processed by independent sample t-tests and Levene’s variance homogeneity tests. The level of significance was set at 0.05. All of the statistical analyses were performed using SPSS 20.0.0 (SPSS Inc., 2009; Chicago, IL, United States).

The calculated sample size was 14 (power: 90%). Finally, there were 17 IKD patients included in this study (Table 1). Most of the IKD patients were injured by low-energy trauma (n = 14): three subjects suffered a sprain of knee when jogging or other exercising; four sprained the knee by themselves during farming; two sprained themselves by accidents during indoor daily life; five were caused by motor vehicle accident. The others were caused by high-energy traffic accident (n = 2) or fall (n = 1). All of the patients were transferred by ambulance to the Emergency department, and urgent reduction (stage-1 surgery) was performed in 0–3 days (Table 1). The knee was fixed in a slight flexion and valgus posture with severe ecchymosis on the medial side, and sometimes the “dimple sign” can be observed (Figure 1A). CT showed posterolateral knee dislocation (Figure 1B). None of the patients had vascular (both the posterior tibial artery and dorsalis pedis artery) or peripheral nerve (both the sensory and motor) injury.

The stage-1 surgery was performed in 0.76 ± 0.90 days after injury (Table 1). Two patients had reversible joint stiffness after the stage-1 surgery (Table 1), and they were intervened by physics therapy with extended rehabilitation for 1–2 weeks until an ROM of 0–120° was restored. No infection or DVT was observed.

After the intermediate rehabilitation, all patients had obtained normal knee alignment (Figure 1D), ROM (0–120°), and valgus/varus stability in the pre-stage-2 period (Table 2). MRI showed ACL and PCL ruptures in all patients (Figure 1C). Lachman and posterior drawer test also found notable anterior–posterior instability in most patients (Figures 2E–G).

The duration from stage-1 to stage-2 surgery was 6.94 ± 1.20 weeks (Table 1). Three subjects had hemarthrosis (10–20 ml) with transiently absorptive fever (<38.0°C) (Table 1), but those implications disappeared spontaneously in 1–2 days. Finally, all patients underwent the arthroscopic reconstruction of ACL and PCL with unilateral semitendinous and gracilis tendons as auto-grafts (Figure 2H).

The average follow-up time was 11.18 ± 2.53 months (Table 1). There were three subjects missing at 1 year follow-up, who cannot be contacted through phone call/e-mail or claimed that they cannot participate anymore, and the total missing rate was 17.65% (Table 1). There were five and three subjects who had mild KFC at 0.5 years and 1 year follow-up, respectively (Table 1).No infection or DVT was observed during the follow-ups.

Most of the subjects had achieved normal joint stability during follow-ups (Table 2). The positive rates of Lachman, posterior drawer, pivot shift test, and forward and backward shift decreased significantly at 0.5 years and 1 year follow-up than those at the pre-stage-2 period, and no difference was found between 0.5 years and 1 year follow-up (Table 2). The valgus/varus instability rate showed no statistical difference among the three periods (Table 2).

Every subject had achieved a normal knee ROM at both follow-ups (Table 3). The knee extension angle at 0.5 years follow-up had an increasing trend compared to that in pre-stage-2, and there was no difference between 0.5 years and 1 year follow-up (Table 3). At 1 year follow-up, the knee extension angle showed a significant increase compared to that in pre-stage-2 (Table 3) (Figure 3). The knee flexion angle at 0.5 year follow-up showed a significant increase than that at pre-stage-2, and this angle was further increased at the 1 year follow-up (Table 3). The flexion angle at the 1 year follow-up was much bigger than that at pre-stage-2 (Table 3) (Figure 3). Lysholm and IKDC scores increased significantly at 0.5 years follow-up than those at pre-stage-2, and those parameters were further increased at the 1 year follow-up (Table 3). The normal knee alignment was observed at 0.5 years and 1 year follow-up (Figures 1E, F).