Incidence and risk factors for non-union of the superior ramus osteotomy when hip dysplasia is treated with periacetabular osteotomy

ABSTRACT Periacetabular osteotomy (PAO) is a well-established surgical treatment for hip dysplasia. Few studies report risk factors for the development of superior ramus osteotomy non-union. The purpose of this investigation was to document the incidence and risk factors for this complication. We identified 316 consecutive hips that underwent PAO for symptomatic acetabular dysplasia with a minimum 1-year radiographic follow-up. We developed and validated a technique to measure the superior ramus osteotomy location on anterior-posterior (AP) pelvis radiographs and computed tomography. Logistic regression with generalized estimating equations was used to evaluate the relationships between odds of non-union and potential demographic and radiographic predictor variables in univariate and multivariate analyses. Twenty-nine (9.2%) hips developed superior ramus non-union. Age {median [interquartile range (IQR)] 23 years (18–35) healed versus 35 years (26–40) non-united, P = 0.001}, pre-operative lateral center-edge angle (LCEA) [16° (11–20) healed versus 10° (6–13) non-united, P < 0.001] and the distance from the superior ramus osteotomy to the ilioishial line [15.8 mm (13.2–18.7) healed versus 18.1 mm (16.2–20.5) non-united, P < 0.001] varied significantly between groups. Using multivariate analysis, moderate-to-severe dysplasia [LCEA < 15°, odds ratio (OR) 5.95, standard error (SE) 3.32, 95% confidence interval (CI) 1.99–17.79, P = 0.001], increased age (5-year increase, OR 1.29, SE 3.32, 95% CI 1.105–1.60, P-value = 0.018) and distance from the ilioishial line (3-mm increase, OR 1.67, SE 0.22, 95% CI 1.29–2.18, P < 0.001) were at increased risk of developing non-union. Superior ramus osteotomy non-union is common after PAO. Older age, moderate-to-severe dysplasia, and more medial osteotomy location were independent risk factors for non-union. Consideration should be made in high-risk patients for a more lateral superior ramus osteotomy and adjuvant medical and surgical interventions.

Further investigation is needed into the incidence and modifiable risk factors for superior ramus non-union when hip dysplasia is treated with PAO.It has been suggested that higher body mass index (BMI), older age, and more severe dysplasia increase the risk of non-union [12][13][14][15][16].The impact of osteotomy location on the superior ramus has not been investigated as a risk factor for non-union.With the majority of the literature reporting the risk for superior ramus non-union as >5% [12][13][14], documenting modifiable risk factors for non-union would guide surgical technique and identify patients who may benefit from an adjuvant intervention to facilitate healing.
The purpose of this investigation was to (i) determine the incidence of superior pubic ramus non-union at a minimum 1-year radiographic follow-up after PAO to treat hip dysplasia, (ii) identify patient-specific risk factors associated with nonunion of the superior pubic ramus, and (iii) determine if the position of the superior ramus osteotomy influences the development of non-union.We hypothesized that older age at the time of surgery, more severe radiographic acetabular dysplasia, obesity, and more medial superior ramus osteotomy would increase the risk of non-union at a minimum 1-year radiographic follow-up.

PATIEN TS A ND M ETHODS
We retrospectively identified 320 patients who underwent a PAO to treat hip dysplasia from January 2003 to May 2020 with a minimum 1-year clinical and radiographic follow-up.Demographic information including age, sex, and BMI at the time of surgery was recorded.Four hips were excluded from the study for incomplete radiographs and clinical follow-up.The final cohort included 316 hips in 246 patients, 81.7% (n = 201) were women and the median [interquartile range (IQR)] age at the time of surgery was 23.9 (18.0-36.0)years (Table I).PAO was performed by two surgeons (T.O.M. and M.C.W.), using previously described techniques [20], with osteotomies in the same order: superior ramus, incomplete osteotomy of the ischium and supra-acetabular osteotomy followed by osteotomy along the posterior column.The superior ramus osteotomy was performed with a straight stiletto osteotome using a Homan retractor to medially displace the hip flexors.The incomplete ischial cut was performed using a curved chisel, and the supra-acetabular osteotomy was performed using an oscillating saw and a combination of straight and curved osteotomes to complete the osteotomy along the posterior column.The post-operative protocol included flat foot-touch weight-bearing for 6 weeks followed by progression to weight-bearing as tolerated.
Two authors (G.S. and J.H.) independently reviewed the medical records to retrospectively document patient demographics and clinical follow-up.All patients had a pre-operative standing anterior-posterior (AP) pelvis and false profile radiograph to measure the lateral center-edge angle (LCEA) of Wiberg [21] and the anterior center-edge angle (ACEA) of Lequesne [22], respectively.These radiographs were repeated at 6 and 12 months post-operatively.Measurements of LCEA and ACEA were performed on immediate pre-operative and 6 months postoperative radiographs by a single reviewer (M.C.W.).Assessment of superior ramus radiographic union was performed by two reviewers (G.S. and J.H.) on standing AP pelvis radiograph at a minimum of 12 months after PAO surgery.Non-union was defined as non-contiguous bony union with a persistent radiolucent line through the pubic ramus osteotomy.We defined a symptomatic non-union as the presence of radiographic nonunion with persistent anterior or midline pelvis pain documented clinically.On the immediate post-operative AP pelvis radiograph, we measured the distance from the superior ramus osteotomy to the ilioischial line by first, drawing a line superimposed over the ilioischial line and then a line parallel to this from the tip of the superior-medial aspect of the osteotomy.We then measured the distance between these two lines (Fig. 1).Out of 316 hips, 260 (82%) hips underwent an immediate postoperative pelvis computed tomography (CT) after PAO.Using the post-operative CT scan, VITREA® (Canon Medical Informatics, Inc.) software allowed for the creation of oblique reconstructions to measure the distance from the medial wall of the acetabulum to the superior ramus osteotomy (Fig. 2).These measurements of the osteotomy position on the superior ramus and evaluation of superior ramus osteotomy union were repeated in 20 consecutive hips by 3 reviewers (M.C.W., J.D. and G.S.) to evaluate inter-rater reliability.

Statistical analysis
Statistical analysis was performed using the SAS software version 9.4 (SAS institute Inc., Cary, NC, USA) and RStudio software version 4.0.2(clusrank procedure [23]).The intraclass correlation coefficient (ICC) was used to describe the inter-rater reliability for osteotomy position on the superior ramus, while Kendall's coefficient of concordance was used to describe the inter-rater reliability for superior ramus osteotomy union.Participant and hip characteristics were described overall and by the incidence of non-union.Categorical variables were described as frequency (percentage), and chi-square tests were used for non-union group comparisons.All continuous variables were reported as median (IQR) due to non-normal distributions, and the Wilcoxon rank-sum test for clustered data was used for nonunion group comparisons.Logistic regression with generalized estimating equations to account for correlation between hips in participants with bilateral osteotomies was used to model the relationship between odds of non-union and potential predictor variables in univariate analyses.These variables included age, BMI, pre-operative and post-operative LCEA, distance of the superior ramus osteotomy from the ilioischial line and distance to the osteotomy from the joint line.The final multivariable model was selected using the quasi-information criterion and included variables significantly associated with odds of non-union.A P-value <0.05 was considered statistically significant.

R E SULTS
Twenty-nine of 316 hips (9.2%) were non-united at the superior ramus osteotomy on 12-month post-operative standing AP pelvis radiographs.Ten (3.2%) hips were documented as symptomatic with persistent anterior or midline pelvis pain.Six out of the 29 hips with superior ramus non-union developed an ischial stress fracture presumed to be a sequela of the superior ramus non-union.Five patients with superior ramus non-union and ischial stress fracture underwent a secondary operation for bone grafting and fixation of the superior ramus non-union (1.5%).One patient with superior ramus non-union and ischial stress fracture had resolution of ischium pain with presumed healing of the stress fracture, although the superior ramus remained radiographically non-united.This patient did not undergo a secondary surgery.
Two patients with superior ramus osteotomy non-union underwent fixation and bone grafting through a modified Smith-Petersen approach (same approach used for the PAO surgery), and three patients underwent fixation and bone grafting through a modified Stoppa approach.Two of the patients who underwent bone grafting and fixation of the superior ramus non-union also underwent screw fixation of an ischial stress fracture (Fig. 3).Non-union fixation and bone grafting was performed a range of 14-18 months after index PAO surgery.All patients who underwent bone grafting and revision fixation radiographically healed the osteotomy, but three out of five patients had persistent dysfunction likely due to organ failure with osteoarthritis.Two patients with fixation and bone grafting of the superior ramus osteotomy underwent total hip arthroplasty (THA) at 1 and 6 years after the revision surgery.Another patient reported persistent, significant hip dysfunction but had not undergone THA 2 years after the revision surgery [modified Harris Hip Score (mHHS) 62 and International Hip Outcome Tool score (iHOT) 28.2].Two patients reported good/excellent hip function after revision surgery (one patient 2 years after revision surgery reported an mHHS of 96 and an iHOT score of 84.5, and another patient 17 years after revision surgery reported an mHHS of 84 and an iHOT score of 97.5).
The distance measurement technique for the superior ramus osteotomy had excellent inter-rater reliability for a single reviewer [inter-rater ICC 0.97, 95% confidence interval (CI) = 0.94-0.98,P < 0.0001].The Kendall's coefficient for the presence of superior ramus non-union was 0.91 for a single reviewer.

DISCUSSION
Non-union of the superior pubic ramus is a recognized, but under-reported complication when hip dysplasia is treated with PAO [12][13][14].Although multiple studies report the prevalence of non-union, there are few that elucidate the risk factors.This retrospective study found a 9.2% incidence of non-union at the superior ramus osteotomy in 316 hips that underwent PAO.Increased age, increasing severity of dysplasia and the distance of the superior ramus osteotomy from the ilioischial line on the AP pelvis radiograph are independently associated with non-union 1 year post-operatively.The medial location of the osteotomy on AP pelvis radiograph was associated with non-union, but a comparable measurement on immediate post-operative CT was not significantly associated with non-union, indicating that CT is not necessary to assess risk of this complication.Quantifying the risk of superior ramus non-union after PAO allows a surgeon to identify high-risk patients so that interventions such as vitamin D optimization, nutrition supplementation and supplemental fixation/bone grafting may be implemented in high-risk patients with the goal of reducing this complication.
The incidence of superior ramus non-union of 9.2% is concordant with the reported rates of non-union in previous studies by Selberg et al. [13] (4.1%), Peters et al. [11] (12%) and Biedermann et al. [24] (13%).We also found a 2% (6/316) incidence of ischial stress fracture which is similar to other studies by Espinosa et al. [18] (0.5%), Peters et al. [11] (1.2%), Tsuboi et al. [25] (2.9%) and Hamai et al. [17] (4.7%).All cases of ischial stress fracture in our series were associated superior ramus non-union.Two of the five patients who underwent a procedure for fixation of the non-union also had screw fixation of the ischial stress fracture with subsequent healing of the stress fracture.In the native pelvis, higher forces are transmitted through the superior pubic ramus compared to the inferior pubic ramus [12].This is altered by superior ramus non-union, and the additional stress can result in an ischial stress fracture.
Our study found that older age and more severe dysplasia independently predicted superior ramus osteotomy non-union after PAO.A recent study reporting the incidence of all types of non-union after PAO found that non-union was much more common 6 months after surgery (55%) [13].Factors that predicted non-union at 6 months included age, obesity and severity of dysplasia (measured with LCEA).By 12 months after surgery, the incidence of non-union was down to 8% indicating that healing progresses >6 months after surgery.At 12-month followup, only age was an independent predictor of non-union.In our series with 12-month follow-up, the median age of patients who developed non-union was 35 years compared to 23 years in patients who healed the osteotomy (P = 0.0010).Our study also confirmed that more severe dysplasia (also quantified with LCEA) increased the risk for development of a superior ramus non-union (pre-operative LCEA median 16 ∘ healed versus 10 ∘ non-united, P < 0.0001).This is not surprising given the large displacement of the ramus that is required to achieve adequate coverage in severely dysplastic hips.Although it may seem inevitable that the ramus develops onto non-union in some cases with a large correction, the ramus appears to be capable of bridging significant gaps in young patients (Fig. 4).In other literature, higher BMI was found to be a risk factor for the surgical complications [15,16], including non-union.Novias et al. [16] reported 7 of 215 PAO surgeries with non-union in the nonobese cohort compared to 5 of 65 cases of non-union in the obese cohort.We did not find increased BMI to be significantly associated with superior ramus non-union in our series.However, the median BMI of patients who went on to develop a non-union was 29.1 kg/m 2 compared to 25.1 kg/m 2 (P = 0.0697) for patients who healed the osteotomy.Increased BMI does put the patient at an increased risk for complications and worse post-operative outcomes as demonstrated in previous studies [15,16].Hence, it is prudent to counsel patients who are older, have a higher BMI and have severe dysplasia that they are at increased risk for complications with PAO, such as a non-union.
Previous studies have postulated that a more medial superior ramus osteotomy has a higher risk for non-union as the medial superior ramus bone is narrower with less contact area Fig. 4. Pre-operative standing AP pelvis of a 24-year-old female who underwent PAO for hip dysplasia with subsequent 3-week post-operative standing AP pelvis radiograph and 1-year follow-up standing AP pelvis radiograph demonstrating the significant ability to heal a medial superior ramus osteotomy with significant displacement in a young, healthy individual.
Non-union after PAO • 85 and requires more displacement for adequate correction [12].Our study found a significant difference in the distance measured from the ilioischial line to the superior ramus osteotomy on standing AP pelvis between united and non-united osteotomies (15.8 mm compared to 18.8 mm for a hip with non-union, P = 0.0009).When a surgeon performs a PAO for a patient with other known risk factors such as older age and moderate-tosevere dysplasia, a more lateral superior ramus osteotomy may be beneficial for healing.This finding is concordant with another study by Matsunaga et al., who found that a gap >5.1 mm at the superior ramus osteotomy was an independent risk factor for a non-union [14].On CT, we did not find that union was associated with medial distance of the osteotomy from the medial wall of the acetabulum.Although the difference was not significant, the median distance of the osteotomy from the joint in the healed group was 18.3 compared to 21.1 mm for the non-union group, which approached statistical significance (P = 0.052).These radiographic measurements indicate that location the superior ramus osteotomy is important for healing and should be considered during PAO surgery.
Although fixation and bone grafting of the superior ramus non-union resulted in radiographic union in all revision cases, three out of five hips had persistent dysfunction (two converted to THA and one had significant impairment indicated by patientreported outcomes).Failure of these hips was attributed to osteoarthritis progression.When considering revision surgery for superior ramus non-union, the status of osteoarthritis progression should be carefully assessed.Diagnostic intra-articular hip joint injection may be considered to determine if the source of persistent pain originates from the non-union or the hip joint.Regardless the patient should be informed of concerns about persistent hip dysfunction despite intervention.
Fixation of the ischium stress fracture was performed in two revision surgeries.Although this is technically achievable with a more medial ischial stress fracture as shown in Fig. 3, this may not be necessary as union of the superior ramus indirectly adds stability for the ischial stress fracture to heal.

Limitations
This is a retrospective single-center study of two-surgeon's practice over 17 years.These findings may not be broadly applicable to other centers and surgeons.Additionally, since this study included cases performed early in the experience by each surgeon, the frequency of the complication may be overestimated because some surgeries were performed early in the learning curve.Also, the sample of hips with a non-union of the superior pubic ramus is small (29) in relation to the study population (316), and hence, this may limit our ability to identify further risk factors.Finally, a significant proportion did not have a postoperative CT (56 of 316 hips).This may have limited our ability to find statistical significance with respect to non-union rates and the CT measurement.

Conclusion
We found that non-union of the superior pubic ramus is a common complication following PAO and is associated with several independent risk factors; older age at time of surgery, moderateto-severe dysplasia and medial position of the superior ramus osteotomy.This study will allow surgeons to quantify risks of surgery and to help devise strategies to reduce complications preoperatively including optimization of bone density, vitamin D status, nutrition and weight loss.Intra-operatively, surgeons may choose to augment healing with a bone graft and/or additional fixation, lateralize the superior ramus osteotomy and prolong the duration of flat foot-touch weight-bearing in high-risk patients.

Fig. 1 .
Fig. 1.Method for the measurement of the distance of the superior ramus osteotomy from the ilioischial line on the AP pelvis radiograph: the first line is drawn along the ilioischial line.The second line is drawn parallel to ilioischial line to the superior tip of the intact medial ramus.The distance between the two lines is measured (22.98 mm in this case example).

Fig. 2 .
Fig. 2. CT of the same patient in Fig. 1 in the immediate post-operative period showing the distance of the osteotomy site from the hip joint (29.2 mm).

Fig. 3 .
Fig. 3. Pre-operative radiograph of a 38-year-old female with symptomatic hip dysplasia.Radiograph of the left hip at 1-year post-operative showing a non-union of the superior pubic ramus and a stress fracture of the inferior ramus.Radiograph 6 months after revision fixation and bone grafting of the non-union site and screw fixation of the inferior ramus showing a healed non-union with clinical resolution of symptoms.Reported an mHHS of 96 and an iHOT score of 84.5, two years after revision surgery.

Table I . Demographics with pre-and post-operative radiographic measurements in the entire cohort, hips with united superior ramus and hips with non-united superior ramus
a Median (IQR), b Number (percent), P-values are for comparisons between healed and non-united hips.