Introduction

Medial opening-wedge high tibial osteotomy (MOWHTO) is a well-established treatment option for medial compartment osteoarthritis (OA) or osteonecrosis with varus malalignment [1,2,3,4,5]. In properly indicated patients, clinical evidence shows that decreased pain and elevated function scores can be achieved by shifting the mechanical axis to a position lateral to the midpoint of the tibial plateau [6, 7]. Nevertheless, previous studies have reported complications, including lateral hinge fractures (LHF), increase in posterior tibial slopes and patellofemoral joint pressure, and alteration of patella heights following otherwise successful MOWHTO procedure [8,9,10,11,12,13,14,15,16]. Among the complications of MOWHTO procedures, LHFs are reported to be the main cause of instability, further displacement (which leads to serious further complications, such as malunion or nonunion), and loss of correction [17,18,19,20,21].

The incidence of LHFs has been correlated with the osteotomy aiming point and sufficiency of the osteotomy [22, 23]. However, a correlation between the opening distance of the osteotomy and the occurrence of LHFs is controversial [22, 24, 25]. Takeuchi et al. [21] introduced a system to classify three types of lateral hinge fractures according to the position of the fracture relative to the proximal tibiofibular joint. Among the three types of fractures, type II and III were considered unstable and were most related to nonunion or malunion of the fracture [21, 26].

Acknowledging the importance of LHFs in patients’ clinical courses, many researchers have focused on surgical tips to prevent such complications and have suggested using techniques, such as optimal aiming point of the osteotomy and sufficiency of the osteotomy, to reduce LHFs [21,22,23, 26, 27]. However, the incidence of LHFs following these previously suggested surgical tips has not been reported. Furthermore, there are only limited reports regarding the clinical outcomes of LHFs occurring during MOWHTO procedures, compared to those without LHFs, using locking compression plates.

Therefore, the aim of this study was to assess the incidence of LHFs during MOWHTO procedures and to further analyze any decrease in the incidence using previously suggested surgical tips and accumulating surgical experience. We also compared whether patients with LHFs had radiographically and clinically inferior outcomes compared to patients with no LHFs after 1 year. We hypothesized that the incidence of LHFs would be reduced over every 30 cases as surgeons became accustomed to the procedure and that patients with LHFs would show radiographic and clinical outcomes similar to those of patients without LHFs.

Methods

This retrospective cohort study was approved by the Institutional Review Board of our hospital. All patients provided written informed consent. Between July 2013 and January 2017, a total of 143 MOWHTOs were performed in 129 patients by a single surgeon at our institution. The indication for MOWHTO was symptomatic medial unicompartmental knee OA with varus malalignment in patients less than 65 years of age. Varus malalignment of 5° or more in the femorotibial angle induced by an abnormal mechanical medial proximal tibial angle (MPTA) was considered an optimal indication. Patients with symptomatic lateral compartments; patellofemoral OA; active inflammatory arthritis, such as rheumatoid arthritis; flexion contractures of more than 15°; or range of motions below 120° did not meet our indication criteria for MOWHTO. Patients who underwent MOWHTOs due to post-traumatic arthritis, had fixating devices other than locking plates used, or had concomitant ligament surgery were excluded from the study. Retrospective data were collected from patients with a minimum follow-up period of 1 year. Eight patients had MOWHTOs for reasons other than primary OA, such as osteonecrosis (one patient) and post-traumatic arthritis (seven patients). A total of three patients were lost to follow-up. Ultimately, 132 MOWHTOs in 121 patients were retrospectively included in the study. To evaluate the trends of incidence of LHFs over time, all 132 MOWHTO cases were chronologically divided into four groups. Each group consisted of an equal 33 cases. The cases were further divided into two groups depending on the presence or absence of LHFs during the MOWHTO procedures.

Patient demographics and comorbidities were evaluated preoperatively, including body mass index (BMI), hypertension, diabetes mellitus, anemia (Hb < 12.0 g/dL), and smoking and drinking statuses. Radiographic evaluations, including hip–knee–ankle angles (HKA angle), MPTAs, weight-bearing line (WBL) ratios, and posterior tibial slope (PTS) angles were performed preoperatively, immediately postoperatively, and at 3 months, 6 months, and 1 year postoperatively, as well as annually thereafter. The HKA angle was assessed on a weight-bearing full-length hip-to-ankle anterior–posterior (AP) radiograph, by drawing the femoral mechanical axis from the center of the femoral head to the midpoint of the tibial spines, and the tibial mechanical axis from the midpoint of the tibial spines to the center of the talar dome [28]. If the HKA angle pointed to varus, the value was described as negative, if the angle pointed to valgus, the value was described as a positive value. The MPTA was also measured on a weight-bearing full-length hip-to-ankle AP radiograph using the tibial mechanical axis and the medial articular surface of the proximal tibia [28]. Radiologic measurement of the WBL ratio was conducted by calculating the proportion of the contact points, starting from the medial edge of the tibial plateau, using the mechanical axis from the femoral head center to the ankle joint center [29]. The PTS was measured using a posterior tibial inclination and a right angle of the tibia anatomical axis bisecting the tibia shaft on true lateral view of the knee.

The osteotomy angle, defined as the angle measured in the knee AP radiograph by lines between the perpendicular line of the tibial long axis bisecting line and the Kirschner wire as a guide for the primary osteotomy, was compared between the groups (Fig. 1). Additionally, the rates of achieving the so-called “safe zone” at the tip of the Kirschner wire, between the fibular head tip and the circumference line of the fibular head, were compared between the groups [22]. Postoperative computed tomography (CT) scans were routinely performed 2 days post-surgery to detect LHFs missing on the immediate postoperative simple radiographs. Clinical evaluations were done using the Knee Society Scores (KSS) [30] and the Western Ontario and McMaster Universities Osteoarthritis Index(WOMAC) scores [31] preoperatively, at 3 months, 6 months, and 1 year postoperatively, as well as annually thereafter.

Fig. 1
figure 1

Osteotomy angle was measured on knee anterior–posterior view using perpendicular line of tibia shaft bisecting line and Kirschner wire as a guide for primary osteotome

Full size image

LHFs were confirmed by definite cortical disruptions seen on postoperative simple radiographs or CT scans. Every case of a confirmed LHF was then classified using the criteria of Takeuchi et al. [21]. A type I fracture had a fracture line that extended just proximal or within the tibiofibular joint, type II was defined as a fracture with a fracture line that extended distally to the tibiofibular joint, and type III was defined as a fracture with a fracture line that extended to the lateral plateau of the proximal tibia.

Evidence of bony union at the osteotomy site was confirmed by an increase in density due to bone bridges. Therefore, the union of a MOWHTO was assessed based on the osteotomy filling index introduced by Brosset et al. using an AP radiograph [32]. From lateral to medial opening wedge, areas were divided into five zones. Each zone corresponded to 20% of the total width. If the filling or bridging of the callus extended into 40 to 60% of the area (zone 3), the osteotomy site was considered stable and united.

Radiographic assessments were conducted by two blinded independent orthopedic surgeons. The average value from these two independent observers was used. The intraclass correlation coefficient (ICC) values of the interobserver and intraobserver reliabilities were substantial to nearly perfect (kappa > 0.80).

Surgical technique

Preoperative measurements were made by an orthopedic surgeon participating in the operation. A single senior surgeon double-checked the preoperative plan before surgery to diminish the opportunity for error. The target of correction was 62.5% of the tibial plateau measured from medial, which is referred to as the Fujisawa point [33]. The correction angle was determined using the Dugdale method, with a weight-bearing full-length hip-to-ankle radiograph [34].

The surgical table was prepared with an additional transparent lower extremity extension board for fluoroscopic identification at the center of the hip joint and in the ankle joint. All patients were positioned in a supine position under general anesthesia. Arthroscopic procedures for the medial compartment, such as microfracture or meniscectomy, were done on demand, as necessary. Subsequently, a 5 cm vertical incision was made on the medial aspect of the proximal leg. The Pes anserinus and the superficial medial collateral ligament were released in all patients to expose the posteromedial crest of the osteotomy site. Under fluoroscopic control, a locking plate (TomoFix®, DePuySynthes, Solothurn, Switzerland) was applied to fix the level of the plate position relative to the joint line and the osteotomy entry. With the use of two Kirschner wires, the endpoint of the osteotomy was targeted toward the “safe zone” [22].

A biplanar osteotomy was carried out in all patients. Primary osteotomies were done on the distal side of the Kirschner wire guides, reaching near the lateral cortex of the tibia, while protecting the neurovascular structures with the use of a transparent Hohmann retractor [35]. A sufficient osteotomy, reaching 5 mm near the lateral tibial cortex, was done on both the anterior and posterior cortices toward the level of the fibular head, to prevent LHFs [23]. With the use of a reciprocating saw, a secondary osteotomy was performed in an ascending manner, just posterior to the tibial tuberosity in all cases. The osteotomy gap opening was meticulously controlled by the angle scale of the spreader. Under fluoroscopic control, the centers of the femoral head and the talar dome were identified and the lower extremity alignment was adjusted until an electrocautery cable was able to pass through the Fujisawa point at the tibial plateau. During this adjustment, a gentle valgus force was applied to minimize the effect of the lateral soft tissue laxity of the knee [36]. After obtaining an ideal correction, the osteotomy was fixed using a Tomofix plate [37]. To promote bony union, allogeneic bone chips grafts (123 cases) or autologous iliac bone grafts (1 smoker case) were performed to fill the osteotomy gap. Eight patients had neither allogeneic nor autologous bone grafts at the osteotomy gap [32].

Postoperative rehabilitation

Regardless of the presence or absence of LHFs, all patients underwent the same postoperative care. On the first postoperative day, quadriceps-setting exercises and continuous passive motion of the knee joint were initiated. A gradual increase in the range-of-motion was instituted, according to individual compliance. Non-weight-bearing three-point crutch ambulation was taught to all patients immediately following surgery. Partial to full weight-bearing was allowed 6 weeks after surgery for all subjects, including LHF patients [36].

Statistical methods

All data were reported as the average and standard deviation. Parametric or non-parametric tests were performed according to the results of tests for normality. A chi-squared test was also performed to compare the incidence of LHFs between the four chronologically divided groups. A Mann–Whitney U test was performed to analyze numerical data. A p value of < 0.05 was considered to be statistically significant. Statistical analyses were performed with SPSS version 22.0 software (SPSS Inc., Chicago, Illinois).

Results

LHFs were detected postoperatively in 32 knees, for an incidence of 24.2% (32/132), in a mean follow-up period of 23.8 months (range 12–48 months). Among the 32 cases with LHFs, 26 were detected on immediate postoperative radiographs and six were diagnosed from postoperative CT scans (Fig. 2). There were six cases of LHFs which were detected by fluoroscopy during the surgery, but additional fixation procedures were not done. Referring to the Takeuchi classifications, 28 cases (87.5%) presented as type I LHFs, where the fracture line extended just proximal to or within the tibiofibular joints, one case (3.1%) was identified as a type II LHF extending distal to the tibiofibular joint, and three cases (9.4%) were type III LHFs which extended to the lateral plateaus of the proximal tibias (Fig. 3).

Fig. 2
figure 2

A 61-year-old female underwent medial opening-wedge high tibial osteotomy. An immediate postoperative radiography of the right knee without definite lateral hinge fracture (a). At postoperative 2 days, CT scan shows lateral hinge fracture (b). Interference of the locking screw with the osteotomy gap was unintended

Full size image
Fig. 3
figure 3

A 39-year-old male underwent medial opening-wedge high tibial osteotomy, Takeuchi type I fracture was noticed on immediate postoperative radiography (a). A 53-year-old female with Takeuchi type II fracture (b). A 63-year-old female with Takeuchi type III fracture (c). All fractures healed uneventfully

Full size image

In terms of demographics, there were no statistical differences in age, gender, BMI, smoking or drinking status, underlying diabetes, hypertension, or anemia between the groups. The only difference between the groups was the operated side (Table 1).

Table 1 Demographic data and preoperative clinical status
Full size table

One hundred thirty-two knees with LHFs were arbitrarily divided into four groups of the same size in chronological order, as shown in Fig. 4. Of the first 33 patients, seven had LHFs. This pattern continued as nine patients among the second 33 patients, eight patients among the third 33 patients, and eight patients of the last 33 patients had LHFs. The incidence of LHFs was determined to be 21.2%, 27.3%, 24.2%, and 24.2% in each sequentially divided group over time, and were not statistically significant (p = 0.954).

Fig. 4
figure 4

Incidence of LHFs among 132 MOWHTOs divided into four groups in chronological order with 33 cases in each group. LHF lateral hinge fracture. *Incidence of LHF in each group

Full size image

Radiographic comparisons between the LHF and non-LHF groups are shown in Table 2. The osteotomy angle in both groups was not statistically different (p = 0.285). The mean osteotomy opening distance was not statistically different between the groups (p = 0.134). There were no statistic differences in the changes in HKA angles, MPTAs, and WBL ratios 1 year post-surgery between the groups (p > 0.05 in each case). However, the non-LHF group showed significantly faster bony unions than the LHF group (p < 0.001). Among the 132 cases of MOWHTOs, 123 cases (93.2%) received allogeneic bone grafts, one case (0.8%) received an autogenic bone graft, and eight cases received no grafting (6.1%). The gap filling methods did not have a statistical correlation with the incidence of LHFs (p = 0.612).

Table 2 Preoperative and postoperative radiologic findings
Full size table

Clinical symptom assessment by the KSS and IKDC scores revealed statistically significant improvements in both the non-LHF and LHF groups compared to baseline scores (Table 3). And, no statistical difference in KSS and IKDC was shown between the groups after 1 year (p > 0.05 in each case). Five cases in the non-LHF group showed wound infections, including four cases of superficial infections which were controlled by the use of intravenous antibiotics in two cases and incisional drainage of the infection site with intravenous antibiotics in the other two cases. Deep infection occurred in one knee, which had undergone early removal of hardware 10 months postoperatively, with evidence of union of the osteotomy site. Symptomatic deep vein thrombosis was seen in one case from the non-LHF group and was resolved with conservative care.

Table 3 Preoperative and postoperative clinical scores
Full size table

Discussion

The most important finding of this study was that there was no change in the incidence of LHFs during MOWHTOs over time. The results might indicate that surgical tips reported by others did not improve the learning curve enough to prevent the occurrence of LHFs during MOWHTOs. The incidence of LHFs was maintained at a constant rate over time. However, patients with or without LHFs showed similar clinical and radiographic results, except for time to bony union.

The occurrence of LHFs as a consequence of MOWHTOs has been an important issue. Furthermore, LHF can be a critical factor in correction loss, instability, and delayed union or nonunion [17,18,19,20, 38, 39]. In a mechanical study, Miller et al. [19] reviewed 50 replicate tibias to evaluate the stability of LHFs during MOWHTOs with non-locking osteotomy plates. In their report, disruption of the lateral cortex resulted in a reduction of axial and torsional stiffness, subsequently causing instability of the osteotomy site, which may contribute to delayed union or nonunion. van Raaij et al. [39] reported 15 cases of LHFs from 43 MOWHTOs in a prospective cohort study using Puddu plates. At the 1-year follow-up, the MOWHTO cases with LHFs had correction loss following recurrent varus malalignment, and two cases showed nonunion, requiring additional surgeries.

Nevertheless, according to our findings, LHFs did not result in considerable loss of correction or nonunion compared to patients without LHFs. In this study, particular attention was paid to technical advancements in MOWHTOs reported by many authors. Such techniques included aiming the osteotomy at an area between the tip of the circumference line of the fibular head referred to as the “safe zone” [22], performing a sufficient osteotomy to reach the “lateral zone” beyond the fibular head on the axial plane [23], and avoiding close proximity to the joint when approaching the target point [35]. Type II fractures have especially been reported to lead to a higher incidence of failure, including delayed union, nonunion, and loss of correction [21]. In a review of 71 patients with 82 MOWHTOs fixated with locking plates, Ogawa et al. [23] reported no cases of type II fractures, which is consistent with the fact that no osteotomies were performed below the lower endpoint of the fibula head. Similarly, in our case, among 132 MOWHTOs, only one case showed a type II fracture, and no cases had their osteotomy endpoint distal to the lower margin of the fibula head.

Introduction of a new locking plate specifically designed for high tibial osteotomy fixation has been reported to show favorable results compared to conventional plates [40]. Locking compression plates, as replacements for pre-existing fixation devices, have shown much more firm fixation and resistance to micromotion of the fracture sites [41], which is a strong advantage when LHFs do occur. And, despite efforts to prevent LHFs by targeting the “safe zone” during osteotomies, unintended LHFs can still gain stability through a femorotibial lateral capsule positioned laterally to the fracture site [22].

Attempts to reduce the risk of LHFs in MOWHTOs have been mentioned in many studies [22, 23, 35]. To the best of our knowledge, however, the learning curve of the procedure and the incidence of LHFs are not well-studied. Cheng et al. [42] reported that the complication rates in total knee arthroplasties tended to decrease to a plateau, along with the learning curve, as surgical volume increased over time. We hypothesized that, as surgeons became accustomed to the procedure, the likelihood of LHFs would be reduced. This study included very early cases of MOWHTOs using locking plates by the senior surgeon. Before the use of MOWHTOs, the surgeon had performed lateral closing-wedge high tibial osteotomies with staples for medial unicompartmental OA knees with varus deformities. All cases were required to be performed using techniques reported to prevent lateral hinge fractures. However, in our study, acclimatizing to the procedure did not reduce the risk of lateral hinge fractures. From the earliest to the latest group, the incidence of LHF did not show any expected decrease in LHFs, comparing each sequentially divided group over time.

While LHFs in MOWHTOs seemed not to crucially affect patients’ 1 year radiographic and clinical outcomes, they still represented a drawback of the procedure. The amount of time to union in the LHF group was 7.3 months, compared to 5.0 months in the non-LHF group, showing a statistically significant difference (p < 0.001). Brosset et al. [32] reported union progress in 51 MOWHTO cases using Tomofix without any filling of the osteotomy gap. The average reported period to gap union was 4.5 months, which is consistent with our results from the non-LHF group. Siboni et al. [43] reported a correlation between LHFs and a consolidation delay of the osteotomies in 41 cases of MOWHTOs using locking plates. Of these 41 cases, five showed nonunion, requiring corticocancellous iliac grafts. Nonetheless, in our cases, there was no nonunion case requiring additional surgery. The delayed union of osteotomies with LHFs was not reflected in any poorer clinical outcomes 1 year after surgery.

Takeuchi et al. [21] previously reported classification of LHFs according to the location of the extension of the fracture line related to the proximal tibiofibular joint. In our cases, the majority of the fractures occurred as type I fractures, which is consistent with the report by Takeuchi et al. While previous studies have considered type II and type III fractures to be unstable, the incidence of these fractures were minuscule in our study, which differed from the results of Takeuchi et al. We believe that the occurrence of these type II and type III fractures can be curtailed by following the above-described surgical techniques to prevent LHFs.

The opening distance during the correction has been reported to correlate with the incidence of LHFs in MOWHTOs [21, 25, 44, 45]. Miller et al. [25] reviewed 46 patients with MOWHTOs and reported statistically significant associations between the wedge size and complication incidences, especially in both intra- and postoperative fractures (p < 0.0063). Nonetheless, the opening distance of the osteotomy did not have a statistically meaningful correlation with the incidence of LHFs during MOWHTOs in our study (p = 0.134). Nelissen et al. [24] reviewed 49 cases of MOWHTOs and concluded that there was no significant relationship between the size of the wedge and the occurrence of LHFs, which is consistent with the results of our study. Han et al. [22] also indicated that there was no correlation between opening distances and LHFs.

Since there is no definitively determined entry point for primary osteotomy, the osteotomy angle might be ambiguous and controversial. However, considering the design of the locking plate with threaded screw holes, which allows screw to thread to the plate as a fixed angle, the entry point for osteotomy was decided on the level of the long shaft portion of the plate between proximal and distal screw holes. The osteotomy angle was measured to prevent too shallow or too steep of an angle. Notably, in our study, the osteotomy angle in all cases did not go beyond 44.8° maximum and 20.3° minimum, with a mean value of 33.71 ± 5.68°. An excessively acute osteotomy angle targeting the lateral cortex below the fibula head may result in type II fractures [23], and an osteotomy angle toward the joint line may result in type III fractures.

There were several limitations to this study. First, the absolute number of MOWHTO cases was small, although it was meaningful in the sense that all operations were done by a single senior surgeon and all cases were performed with the same locking compressing plate under consistent surgical techniques. Second, most of the patients undergoing MOWHTO in our study were female. However, a higher female incidence of OA in the Korean population has been reported before [46]. Third, this study had a relatively short follow-up period of a minimum of 1 year. Despite the fact that the assessment of LHFs and union of the osteotomy site was fully possible and comprehensive within this period, a longer follow-up period is still needed. Fourth, the study population was rather inhomogeneous in regard to osteotomy gap filling. Fifth, 6 weeks of non-weight-bearing after surgery as our rehabilitation protocol might have affected the result of our study.

Conclusion

The incidence of LHFs during MOWHTOs remained consistent at 24.2%, despite surgeons’ accumulating experience with the procedure over time. Although patients with LHFs required longer times to achieve bony union, they showed radiographic and clinical results similarly favorable to those of patients without LHFs 1 year after surgery.