Comparison between innominate osteotomy and arthrodistraction as a primary treatment for Legg-Calvé-Perthes disease: a prospective controlled trial

Abstract

Purpose

Legg-Calvé-Perthes (LCP) disease is currently managed by mechanical containment of the femoral head in the hip socket. As evidence suggests that hip distraction may offer a new treatment strategy, we used arthrodistraction as a primary treatment for active forms of LCP disease and prospectively compared the results with the Salter innominate osteotomy.

Methods

A total of 54 children, six years or older of both genders with severe forms of LCP disease in the stages of necrosis or revascularisation, were enrolled. Patients were submitted to either Salter innominate osteotomy (n = 28) or hip arthrodistraction (n = 26). Final radiographs were used to evaluate the Mose index, Wiberg angle, extrusion index and the Stulberg et al. classification.

Results

There were no significant differences in gender, age, lateral pillar classification and average follow-up time between the two groups. The osteotomy group progressed without major complications, but children in the joint distraction group experienced episodes of pin tract pain and infection, leading to the early removal of the external device in one case. Two patients developed joint stiffness, treated by physiotherapy or manipulation, and one child developed subluxation of the femoral head. The average time in distraction was 4.44 months (2.53–7.23 months). In the final evaluation the osteotomy group showed better containment of the femoral head. The Mose index and the Stulberg et al. classification were statistically similar between the two groups.

Conclusions

Despite similar final radiological results, arthrodistraction was associated with a higher morbidity. Consequently, we do not recommend hip distraction as a primary treatment for the early stages of LCP disease.

Introduction

The current treatment for Legg-Calvé-Perthes disease focuses on mechanical protection of the femoral head to avoid hip deformity and future joint degeneration. This protective effect is based on maintaining the plastic epiphysis in the acetabulum, which is attempted either by conservative or surgical methods [1–6]. The natural history of the disease and the concept of “joint at risk” have been accepted as conclusive elements for prognosis and treatment options [7]. In recent times, imaging methods, such as scintigraphy and magnetic resonance, have been introduced for patient assessment and represent additional information that enables the best management strategy for each case to be selected [8–11]. Despite such an armamentarium, many cases fail to respond to treatment, often resulting in a non-congruent hip.

More recently, there has been some evidence to suggest that distraction forces applied to the hip joint can offer mechanical protection to the femoral head and promote bone revascularisation [12–17]. Therefore, hip joint distraction may offer an alternative therapeutic option for patients with Legg-Calvé-Perthes disease (otherwise called Perthes’ disease), as it represents an attractive link between mechanical and biological processes. However, this treatment has not been widely accepted, as the literature does not show strong scientific evidence that it is more beneficial for the patient, when compared with traditional methods. Most of the series published to date indicate the benefit of hip distraction in Perthes’ disease patients; but these studies are considered preliminary [12], or the series are too small to guarantee reliable conclusions [15–17].

Consequently, prospective comparative studies are missing in the literature. From 1994 to 2002, patients with severe forms of Legg-Calvé-Perthes disease were primarily treated at our institution either by hip arthrodistraction or pelvic innominate osteotomy using the Salter technique. The aim of this report is to compare the results obtained with both methods.

Methods

Study design

This study was approved by the Institutional Review Board on Ethics and Research of the University of São Paulo Ribeirão Preto Medical School. This was a prospective non-randomised, non-blind, controlled trial conducted in children with active forms of Legg-Calve-Perthes disease (otherwise called Perthes’ disease) who were referred to a tertiary university hospital in Brazil.

At the first evaluation, the diagnosis was confirmed by radiographs for each child (supine anteroposterior and Lauenstein pelvic views). A complete clinical history was obtained and, based on the radiographs, the following parameters were established: stage of the disease (necrosis, revascularisation, remodelling), per cent of the epiphyseal involvement on the lateral view and the presence, or not, of femoral head subluxation. Hip subluxation was specified by a broken Shenton’s line and an increased distance between the acetabulum and the medial intra-articular part of the metaphysis of the femur. For those cases in the revascularisation stage, the lateral pillar classification was applied [18].

Based on this initial evaluation, each patient was classified as having Perthes’ disease in an active stage (necrosis/revascularisation), with or without a congruent hip (hinge abduction) and/or “at risk” femoral head. The risk condition was considered when at least three of the following findings were present: age older than six years, considerable restriction of hip motion, adduction contracture, necrotic area comprising 50 % or more on the profile view, femoral head subluxation [19] and lateral pillar B or C [18]. Children who were six years of age or older, with a congruent joint, cephalic nucleus in necrosis or revascularisation and “at risk” hip were eligible to be included in the study. Children were excluded if they presented with milder forms of the disease and were instead recommended a treatment strategy of swimming, avoidance of other sports and physical activities, and treatment of pain with ibuprofen. These children were also scheduled for regular follow-up and evaluation to monitor the progression of the disease. Other exclusion criteria were advanced stages of the disease, usually with a deformed femoral head, fixed adduction contracture, acetabular deformity and hinge abduction. In these cases, the condition was classified as irreversible and the patients were treated for sequelae.

For patients included in this study, two treatment options were presented to the child’s guardians: hip distraction or pelvic osteotomy. After thoughtful discussion between the guardians and the surgeons, a treatment strategy was devised and written consent was obtained. Operations were performed with the patient under general anaesthesia or spinal anaesthesia on a radiotransparent surgical table top. Before the operation, the hip was examined under an image intensifier for concentric motion. If necessary, arthrography was performed. If retracted, the adductor muscles were stretched, until symmetrical movement was achieved in relation to the opposite hip. The innominate osteotomy was performed according to the technique described by Salter [20]. However, at the surgeons’ discretion, an osteotomy of the pubic superior ramus was also performed to obtain adequate mobilisation of the acetabular fragment and adequate femoral head covering (older patients).

For hip distraction, we used a unilateral hinged arthrodistractor with two Schanz screws inserted into the supra-acetabular region and two screws applied to the femoral diaphysis. Care was taken to make sure the hinge and the centre of motion of the hip in 20° of abduction were coincident. Following this, joint distraction was applied until widening of the articular space was observed under image intensification. During the postoperative period, hip motion and partial weight-bearing were encouraged. At weekly intervals, the distraction was checked and radiographs were taken, if necessary. The hip distractor device was programmed to be removed when ossification of the lateral pillar had occurred. All patients were followed up at least until complete ossification of the femoral head and closure of the growth cartilage. The final radiographs were used to obtain the Wiberg angle, the Mose evaluation of head sphericity [21], the femoral head extrusion index [22] and the Stulberg et al. classification [23].

Statistics

Data analysis was carried out with the software SAS® 9.2 using the PROC MEANS and PROC FREQ (SAS Institute Inc., Cary, NC, USA). Fisher’s exact test was performed to evaluate the association between groups for qualitative variables: gender, lateral pillar and Mose index. Student’s t test was used in the group comparison for the variables: age at operation, extrusion index, follow-up time and Wiberg angle. The p value was set at <0.05.

Results

From 1994 to 2002, 116 patients with 132 affected hips were referred to our institution; 63 hips were treated surgically, and 69 were managed symptomatically. For the operated patients, 28 hips were treated by Salter innominate osteotomy and 27 hips by arthrodistraction. Eight hips were considered as sequelae and were treated by valgus osteotomy (five hips) or Chiari osteotomy (three hips). Figure 1 outlines this allocation of patients.

Fig. 1.

Flow of participants through each study stage

In the pelvic osteotomy series, there were no cases of infection, but the Kirschner wires in one patient caused skin irritation and had to be removed. Another patient was submitted to physiotherapy because of restricted hip motion, with an uneventful recovery. No other complications were reported.

In the arthrodiastasis series, one patient complained of severe pain on the day after the operation, which was alleviated by the release of the excessive distraction. In addition, patients in this series had frequent skin irritation and mild pin tract infections that were treated with local antiseptics or short periods of oral antibiotics, respectively. However, in one case, the treatment was discontinued because of uncontrolled pin tract infection and pin loosening. This patient was later treated with a varus femoral osteotomy, with a good outcome. This case was considered a failure, but it was not included in the final radiographic evaluation.

All of the children in the arthrodistraction group, in the postoperative period, were able to partially bear weight, except in one case, where the child was afraid and refused to walk, despite psychological assistance. Most of the children were able to move the hip in flexion-extension just a few degrees, but to a lesser extent than the degree of movement achieved during the operation. In four cases, there was no movement at all. One patient suffered a broken femoral screw, but it was not replaced because it occurred in the final stage of treatment. In one patient, progressive subluxation of the hip developed as the distraction progressed. Treatment for this child was paused until there was complete revascularisation of the ossific nucleus during the arthrodistraction, following which lateral subluxation was successfully treated with varus osteotomy, with good head centralisation. The distraction device remained implanted for an average of 4.44 months (range 2.53–7.23 months). Two patients presented with hip stiffness after the device was removed: one patient was treated with physiotherapy and the other with joint manipulation under anaesthesia. Both of these patients presented significant improvement in their articular mobility after additional treatment.

For analysis, 26 hips were available in the arthrodistraction group and 28 hips in the Salter group: 19 individuals (73 %) were evaluated at skeletal maturity in the arthrodistraction group and 17 (61 %) patients in the Salter group. Table 1 shows the descriptive analysis of the data. The statistical analysis showed that both groups were comparable in terms of gender (p = 0.1791), age at operation (p = 0.3658; Fig. 2) and follow-up time (p = 0.7806; Fig. 3). All patients were successfully followed up, with an average follow-up time of seven years for both groups. The statistical analysis also showed no significant differences in the lateral pillar classification (p = 0.7345); however, the lateral pillar was not able to be classified in 11 cases from each group, because the ossification nucleus was not in the revascularisation stage for the first radiograph [24]. Both the extrusion index and Wiberg angle showed statistical evidence that there were differences in the averages between the groups [confidence interval (CI) ≠ 0; p < 0.05; Table 2]. The epiphyseal extrusion index, which measures the percentage of the femoral head that is uncovered, demonstrated lower values for the innominate osteotomy group (p value = 0.0327; Fig. 4). The Wiberg angle, which estimates head coverage and centralisation, showed higher values for the Salter group. Together, these results indicate that Salter innominate osteotomy provides a better coverage and centralisation of the femoral head (Fig. 5). For the Mose measurement, which measures the femoral head sphericity or ‘roundness’, no differences were noted between the two groups (Table 3). At the significance level of 5 %, there is statistical evidence that there were no differences between the groups for the Stulberg classification, a measure of long-term outcomes (Table 4).

Table 1.

Descriptive analysis of data

Group	n (group)	Variable	n (var.)	Mean	SD	Min.	Median	Max.
Arthrodistraction	26	Age at OP	26	99.04	16.59	69	101	134
		Extrusion index	25	0.22	0.13	0	0.21	0.44
		Follow-up	26	87.12	33.34	41	86.5	188
		Stulberg	26	3.12	1.11	1	3	5
		Wiberg angle	26	24.77	8.31	9	23	39
Salter	28	Age at OP	28	94.93	16.46	59	92.5	134
		Extrusion index	28	0.15	0.11	0	0.14	0.41
		Follow-up	28	84.28	40.31	16.53	84.78	163.53
		Stulberg	28	2.75	1.21	1	3	5
		Wiberg angle	28	35.29	7.69	20	35	56

OP operation, var. variable, Min. minimum, Max. maximum

Fig. 2.

Box plot: a comparison of age at operation showed no statistical significance between the groups (p value = 0.3652)

Fig. 3.

Box plot: a significant difference in the Wiberg angle was detected between the groups, with higher values for the Salter group (p value < 0.05)

Table 2.

Comparison of the extrusion index and the Wiberg angle between the arthrodistraction and pelvic osteotomy groups

Variable	Diff.	Estimate	SE	CI (95 %)		p value
				Lower	Upper
Extrusion index	A-S	0.07414	0.03376	0.00636	0.1419	0.0327
Wiberg angle	A-S	-10.5165	2.1776	-14.8862	-6.1468	<0.0001

Student’s t test was used to assess significance. Significance was indicated at p < 0.05

Diff. difference between the arthrodiastasis (A) and Salter (S) groups, SE standard error of the mean

Fig. 4.

Box plot: a comparison of the follow-up time showed no statistical significance between the groups (p value = 0.780)

Fig. 5.

Box plot: extrusion index for the arthrodistraction and pelvic osteotomy groups. There was a statistical difference identified between the groups (p value = 0.0327) with a more extruded head in the arthrodistraction group

Table 3.

Groups versus Mose index [21]

Group	Mose			Total	p value
	0	2	>2
Arthrodistraction	3	7	16	26	0.6454
	11.54 %	26.92 %	61.54 %
Salter	6	6	16	28
	21.43 %	21.43 %	57.14 %
Total	9	13	32	54

0 = perfectly round femoral head, 2 = flattening of the femoral head up to 2.0 mm, >2 = femoral head flattening larger than 2.0 mm

Table 4.

Groups versus Stulberg et al. [23] classification

Group	Stulberg et al. classification					Group total
	1	2	3	4	5
A (%)a	2	6	7	9	2	26
	7.69 %	23.08 %	26.92 %	34.62 %	7.69 %
S (%)a	5	8	5	9	1	28
	17.86 %	28.57 %	17.86 %	32.14 %	3.57 %
Total for each classification	7	14	12	18	3

Analysis for arthrodistraction (A) and Salter (S) groups was performed using the Wilcoxon-Mann-Whitney test

^aPercentage of the entire group

Discussion

Legg-Calvé-Perthes disease is one of the most studied disorders in paediatric orthopaedics; but, the condition is still poorly understood, and there is considerable divergence of opinion, mostly in relation to treatment. However, some consensus has been reached for some parameters. First, most authors agree that the condition is more severe in older children, especially after six to eight years [24], in cases of lateral pillar B and C [25] and when there is more than a 50 % involvement of the femoral head [26]. The concept of “head at risk” is applied to differentiate cases that may benefit from treatment from those that should not be treated [19]. Nonspecific treatment is reserved for cases with a benign natural history or for cases that are too severe so that no preventative treatment would be useful [24].

In our study design, we considered the aforementioned criteria so that our cohort was formed by individuals with more severe forms of the disease. Our study was conducted prospectively, but it was not blinded and was non-randomised. Indeed, the choice of treatment was decided by the child’s guardian as, for ethical reasons, we decided not to randomise the trial. We did not offer the option of braces, because the literature does not provide convincing evidence to support their use [27]. Likewise, we did not consider treating patients with varus osteotomy, as it causes too much femoral shortening in older children [28] and pelvic asymmetry. Therefore, our preference was for the various types of pelvic osteotomy, particularly the Salter technique [20]. Surprisingly, many parents chose to treat their children by arthrodistraction, a new method of treatment, not properly tested at the commencement of this research. We believe that one of the reasons for this choice was that it represented less surgical morbidity in comparison with a conventional surgical procedure. Nevertheless, some bias may have occurred because we were interested in the results of the arthrodistraction; therefore, some influence may have existed favouring this method of treatment. However, if such bias existed, it did not significantly influence the formation of the groups as, within the same time span, we collected a similar number of patients of similar age and disease severity.

In our study, hip morphology and head congruency were evaluated by several parameters. The epiphyseal extrusion index describes the percentage of the femoral head that is uncovered [22]. The Wiberg angle estimates the head coverage and centralisation, and the Mose method [21] defines the femoral head sphericity. It is logical that the more of the head that is spherical and well contained, the better the result. Nevertheless, we did not attribute to these parameters the subjective concept of “good, fair and bad” results, because there is no consensus in the literature about this classification. Instead, we used various parameters to compare the results between the two series. Overall, our results showed that hips treated by Salter innominate osteotomy were more covered and less extruded, but showed no difference in shape (Mose index), as compared with hips treated by arthrodistraction. The Stulberg et al. classification [23] is well accepted to establish prognosis and long-term results and showed no difference between the two groups.

Several studies have evaluated the role of arthrodistraction in the treatment of Perthes’ disease [12–17], but with discrete methodologies. Some authors included soft tissue release [16], whereas others used the non-articulated Ilizarov system [15]. Maxwell et al. [12] applied a unilateral articulated external fixator in 15 hips and used a historical series as a control. The authors considered their results as preliminary, with numerous complications, but felt that the method might benefit patients with Perthes’ disease [12]. Our final results, according to the Stulberg et al. classification [23], support the same opinion. However, when comparing both techniques, not just in terms of end outcomes but considering the whole treatment, the innominate osteotomy group presented fewer complications. In fact, the numerous episodes of pin tract infection and pain, the difficulty in carrying an external device that interferes with clothing and social life and convincing the child to use crutches all significantly affected the day-to-day life of the child following surgery. Additionally, albeit less frequently, more severe complications were noted with arthrodistraction, such as cases where reoperation was indicated to correct technical problems with the external device, or changes in treatment; these issues compromised its indication as a primary treatment for Perthes’ disease.

The strength of this investigation is that it was a prospective study that compared two similar populations applying objective methods of patient selection and evaluation. The drawbacks were that the methods of assessment were based solely on radiographic aspects and did not consider the patient’s satisfaction, sports limitation, limb shortening and hip mobility.

Conclusions

In conclusion, despite similar final radiological results, the arthrodistraction process had an increased morbidity when compared with the innominate osteotomy. Consequently, we do not recommend arthrodistraction as a primary treatment for the early stages of Legg-Calvé-Perthes disease.