The study protocol was approved by the local institutional review board, and informed consent was obtained from all patients. A retrospective review of our institutional total joint registry identified forty-six hips (thirty-nine patients) with Crowe type-IV developmental dysplasia that were treated with total hip arthroplasty between 1992 and 2005. The data in the joint registry were prospectively collected. Thirty-seven hips (thirty-two patients) were treated with a simultaneous subtrochanteric shortening femoral osteotomy and total hip arthroplasty. Nine of these hips (in eight patients) had a total hip arthroplasty with a cemented femoral component and were excluded from this study. This left twenty-eight hips in twenty-four patients who were treated with a simultaneous shortening subtrochanteric osteotomy and cementless total hip arthroplasty. During the same time period, 8848 primary total hip joint replacements were performed at our institution; therefore, this cohort represents 0.3% of the total hip arthroplasties performed at our institution in those years. The operations were performed by four orthopaedic surgeons (including D.J.B., M.E.C., and R.T.T.) who made the decisions regarding the need for an osteotomy. Preoperative clinical data, surgical data, and postoperative clinical data were collected prospectively on a routine basis for all patients.
Clinical Data
Chart review was performed at an average of 4.8 years (range, 2.0 to 13.4 years) postoperatively for the twenty-eight hips (twenty-four patients) (Table I). At the time of final follow-up, nineteen patients (79%) returned for clinical and radiographic evaluation. The remaining five patients filled out written questionnaires and had radiographs sent to us for review. Two patients died (at the ages of fifty-five and seventy-seven years) of causes unrelated to the total hip arthroplasty at three and four years, respectively, after the index procedure. Both had had a well-functioning total hip replacement at the last clinical examination prior to death. No patients were lost to follow-up.
At the clinical evaluation, the patients were weighed, the preoperative or postoperative limb-length discrepancy was measured, the patients were observed for a limp, the Harris hip score was calculated, and any prior operations were recorded. Operative data that were collected included the surgical approach, the operative time, the type of fixation of the osteotomy site, the use of bone grafts, the implant sizes, the type of implant fixation, and the use of an acetabular structural graft.
Limb lengths were measured from the anterior superior iliac spine to the level of the medial malleolus with the patient lying supine. Radiographic data included the location of the acetabular component, healing of the osteotomy site, and component stability. All complications were recorded.
The patients included twenty women and four men with a mean age of 47.6 years (range, thirty to seventy-two years) at the time of the total hip arthroplasty. The mean patient weight was 66.4 kg (range, 52 to 98 kg), the mean patient height was 156.3 cm (range, 128 to 169 cm), and the mean patient body mass index was 27.2 kg/m2 (range, 20.3 to 39.8 kg/m2).
Sixteen (57%) of the hips had undergone previous operative treatment, with nine (32%) having had more than one previous operation. Procedures performed in childhood included open reduction in six hips and closed reduction followed by the application of a cast or orthosis in four hips. An isolated intertrochanteric or subtrochanteric femoral osteotomy had been performed previously in ten of the hips, and a pelvic osteotomy had been performed in seven. One of the patients had a contralateral external hemipelvectomy because of a severe injury forty years prior to the total hip arthroplasty. Associated comorbidities included multiple sclerosis in one patient, spondyloepiphyseal dysplasia in one, poliomyelitis in one, bilateral clubfoot in one, and severe thoracolumbar scoliosis in two.
Operative Technique
The operative approach to the hip was posterior in twenty-six hips and anterolateral in two. A variety of femoral implants was used; these included twenty-four extensively porous-coated stems (nine Solution, seven Bantam, four Prodigy, and four Anatomic Medullary Locking [AML] implants; DePuy, Warsaw, Indiana) and four proximally porous-coated modular stems with a fluted distal element (S-ROM; DePuy). The femoral head size was 22 mm in seventeen hips, 28 mm in nine, and 32 mm in two. A porous-coated acetabular component with dome screws was used in all hips (fourteen Trilogy implants, Zimmer, Warsaw, Indiana; twelve Pinnacle implants, DePuy Orthopaedics; one Omnifit PSL implant, Howmedica Osteonics, Allendale, New Jersey; one HGP II implant, Zimmer). The median outside diameter of the acetabular component was 44 mm (range, 42 to 54 mm). The two 32-mm femoral heads were placed in combination with a 54-mm acetabular component, leaving a minimum polyethylene liner thickness of 7.4 mm at the locking mechanism and 7.9 mm at the dome (Pinnacle, DePuy Orthopaedics). Structural femoral head autograft fixed with screws was used in ten of the acetabular reconstructions to improve lateral coverage of the cup. The structural autograft was contoured and fixed to the ilium with one or two screws.
A transverse subtrochanteric osteotomy was used in twenty-seven hips, and a step-cut subtrochanteric osteotomy was used in the remaining hip. The exact site of the shortening osteotomy varied according to the patient's anatomy, but it was generally 8 to 10 cm distal to the tip of the greater trochanter and always distal to the lesser trochanter. Autologous cancellous bone graft was used at the osteotomy site in twenty-three of the twenty-eight hips. Additional stabilization of the osteotomy site was provided by cortical autograft struts cut from the resected femoral bone fragment in twenty hips and by cortical allograft struts in two hips. Electromyographic monitoring of the sciatic nerve was used in seventeen patients, at the discretion of the surgeons. A temporary abduction hip guide brace was used postoperatively by twelve patients, again at the discretion of the surgeons.
For preoperative templating (Figs. 1-A, 1-B, and 1-C), we placed the cup at the true anatomic hip center and placed the template for the femoral component so that the neck cut was approximately 1 cm proximal to the lesser trochanter. If templating suggested that a reduction of the hip would result in limb lengthening of greater than 3 to 4 cm, then we planned to perform a subtrochanteric shortening osteotomy. When appropriate, the length of subtrochanteric bone anticipated to be removed was calculated to equalize the leg lengths and lengthen the limb by no more than 3 to 4 cm.
In most instances, the proximal part of the femur was approached with use of a posterior approach to the hip. After head-neck resection, the proximal part of the femur was prepared for insertion of a cementless femoral component. The subtrochanteric area of the femur was approached by elevating or splitting a short section of the vastus lateralis. Next, a transverse proximal femoral osteotomy was performed, usually about 10 cm distal to the tip of the greater trochanter. The proximal femoral fragment was then translated anteriorly, and the true acetabulum was identified with use of the round ligament and the transverse ligament as anatomic guides. The acetabulum was prepared with use of small reamers, and the acetabular component was then inserted in the anatomic acetabular position. In most instances, sufficient lateral coverage of the socket was obtained with placement of the cup at or near the normal anatomic position and deepening the socket with reaming to the medial wall. When this was not feasible, we preferred to augment the acetabulum with a superolateral bone graft from the patient's own femoral head18.
Attention was then redirected to the femur, where a second, more distal, femoral osteotomy was performed to shorten the femur by the planned amount. A mean of 4 cm (range, 2 to 8 cm) of femoral bone was resected. In some instances, the osteotomized cylinder of bone removed during the shortening was split longitudinally into two or three fragments and attempts were made to maintain the vascularity of the fragments by retaining their soft-tissue attachments. The distal part of the femur, below the site of the shortening osteotomy, was then prepared for the cementless implant.
When a fully porous-coated stem was to be used, the femoral canal was reamed to a diameter of 0.5 mm less than the diameter of the distal component. For fluted modular stems, the distal part of the femoral canal was reamed to a diameter of 0.5 mm less than the major diameter of the distal splines. The proximal and distal fragments were then aligned and were properly rotated on the trial prosthesis, and the hip was reduced. The sciatic nerve was palpated to ensure that it was not excessively taut. If necessary, additional bone was gradually resected in sequence at the subtrochanteric osteotomy site until satisfactory reduction of the hip was achieved without excessive tension on the sciatic nerve and good apposition was achieved at the osteotomy site. The femoral implant was then impacted into the femur, across the osteotomy site, once the rotational alignment of the proximal fragment and the distal fragment was adjusted to allow approximately 15° of anteversion of the femoral component. Rotational and axial stability of the osteotomy site were achieved by the tight fit of the implant in the proximal and distal bone fragments. When available, autogenous cortical strut grafts from the cylinder of femur removed during the subtrochanteric osteotomy were affixed around the subtrochanteric osteotomy site with two cables or wires to enhance the stability of the osteotomy site and potentiate biologic healing (Fig. 1-C). Cancellous bone fragments from the reaming of the patient's acetabulum were packed along the femoral osteotomy site. The hip was then reduced, and the wound was closed.
Postoperatively, the patients walked with toe-touch weight-bearing for approximately eight weeks. An abduction hip guide brace was used at the discretion of the surgeon, with the patient ideally wearing it at all times. Gradually progressive weight-bearing was permitted according to evidence of radiographic union of the osteotomy site, usually starting at eight weeks postoperatively.
Radiographic Analysis
The change in the hip center was measured by determining the vertical distance from the ischial tuberosity to the center of the femoral head on both preoperative and postoperative anteroposterior pelvic radiographs as previously described14. Healing of the osteotomy site was assessed on postoperative radiographs. The criteria for union were the presence of callus at the osteotomy site, cortical continuity of the proximal and distal femoral fragments, and no progressive gapping at the osteotomy site on serial radiographs14. The radiographs were also examined for component migration and the presence of heterotopic ossification, osteolysis, and acetabular polyethylene wear. Heterotopic bone was assessed with the method of Brooker et al.19. Fixation of the femoral component was evaluated according to the radiographic criteria described by Engh et al.20. Loosening of the acetabular component was defined by the presence of progressive radiolucent lines of >2 mm in thickness in all three zones as described by DeLee and Charnley21, broken fixation screws, or migration or another change in the position of the implant.
Statistical Analysis
After confirming normal distributions and equal variances of the data, we used a two-sided paired Student t test to analyze preoperative and postoperative Harris hip scores and limb-length discrepancy, with p < 0.05 considered to indicate significance.
Source of Funding
There was no external funding source for this study.
Clinical Results
The mean Harris hip score improved significantly (p < 0.01) from 43 points (range, 24 to 89 points) preoperatively to 89 points (range, 53 to 96 points) postoperatively (Table I). The mean pain component of the Harris hip score was 15 points (range, 10 to 40 points) preoperatively and 42 points (range, 24 to 44 points) postoperatively, and this difference was significant (p < 0.01). Preoperatively, there was a severe limp in twelve cases, a moderate limp in twelve, and a slight limp in three. (One patient had had a contralateral hemipelvectomy forty years previously.) Postoperatively, there was a moderate limp in four cases, a slight limp in twelve, and no limp in eleven.
The average preoperative limb-length discrepancy in the twenty patients with unilateral dysplasia was 43 mm (range, 25 to 60 mm). The clinically measured postoperative limb-length discrepancy in the whole patient cohort (excepting the patient with one hip) was <1 cm in thirteen patients and between 1 and 2 cm in ten patients. No patient had >2 cm of limb-length discrepancy (Table I).
Radiographic Analysis
Twenty-six (93%) of the twenty-eight osteotomy sites healed without complications. The two patients with nonunion of the osteotomy site are described below. All twenty-five femoral components that remained in situ after the index arthroplasty demonstrated radiographic evidence of bone ingrowth and stability at the time of the last follow-up. All twenty-seven acetabular components that remained in situ after the index arthroplasty were seen to be stable radiographically at the time of the last follow-up. No osteolysis was observed on follow-up radiographs of any patient. Six patients had asymptomatic Brooker class-I heterotopic ossification seen on follow-up radiographs. The average hip center was translated distally 58 mm (range, 26 to 97 mm).
Complications and Reoperations
An intraoperative femoral fracture occurred in five hips (18%) (Table II). All fractures occurred in the distal femoral segment during either bone preparation or insertion of the femoral component. All fractures were nondisplaced cracks and were treated with a cerclage cable or wire; there were no sequelae. Four patients had postoperative dislocation. In three patients, the dislocation occurred once and did not recur. The fourth patient, who had had an anterior approach, sustained a fracture of the greater trochanter during a fall and had a subsequent nonunion of the trochanter with recurrent dislocation. No clinical evidence of postoperative neurologic injury was observed in any patient. There were also no postoperative infections.
Six hips underwent at least one reoperation, and one of them had two reoperations. One patient had a nonunion of a femoral osteotomy site associated with loosening of the femoral component. It was treated with revision of the femoral stem to a fully porous-coated stem and bone-grafting of the nonunion site at ten months after the index arthroplasty. One patient had a nonunion of the femoral osteotomy site associated with a fracture of the femoral stem at twenty-one months after the index arthroplasty. It was treated with revision of the 10.5-mm extensively porous-coated stem to a 13.5-mm stem with bone-grafting of the osteotomy site. In both instances, the osteotomy site healed uneventfully, and the new femoral component was stable. In one patient, aseptic loosening of an extensively porous-coated femoral component developed in association with a healed osteotomy site. It was revised successfully to a larger femoral component at three years after the index arthroplasty. One patient had aseptic loosening of the acetabular component at seven years after the index procedure, and the cup was revised. Later, the same patient sustained a fracture of the greater trochanter, which was treated with internal fixation. In one patient, the acetabular liner disengaged at six years after the index procedure and a femoral head and liner exchange was performed. Finally, one patient (included among the four with postoperative dislocation noted above) presented at the six-week postoperative visit with dislocation of unknown duration and was treated with open reduction without further sequelae.
To our knowledge, this is the largest reported series in which the results of exclusively cementless total hip arthroplasties with a concomitant femoral subtrochanteric shortening osteotomy were evaluated in patients with Crowe type-IV dysplasia of the hip. Other published series of patients treated with a subtrochanteric shortening osteotomy have included cemented components11-15 or combinations of other techniques17.
The majority of the patients in this series had an improvement of the preoperative limp, but in sixteen cases there was at least a slight limp at the time of the last follow-up. This functional result is not as good as it is in patients treated with routine primary total hip arthroplasty for osteoarthritis, probably because of longstanding abductor muscle deconditioning and gait abnormalities18. However, it represents a substantial improvement compared with the functional results that were achieved with earlier methods of treatment for this complex problem22.
In their series of total hip replacements performed for treatment of congenital dislocation and dysplasia of the hip, Crowe et al. reported an excellent clinical outcome in eleven hips (35%) and a good outcome in sixteen (52%)1. In the series reported on by Numair et al.4, only twenty-five hips (54%) had normal or nearly normal function postoperatively, and Dunn and Hess23 reported that ten of sixteen patients had a persistent abductor lurch. In our series, the limb-length discrepancies were significantly decreased compared with the preoperative discrepancies. Our results are comparable with recent findings following iliofemoral distraction with use of an external fixator22, and similar to those after subtrochanteric osteotomy14,24.
There are several advantages to using a subtrochanteric osteotomy rather than an intertrochanteric osteotomy. It allows simultaneous shortening and correction of rotational abnormalities as well as preservation of the proximal femoral metaphysis. The preservation of the proximal femoral metaphysis facilitates the placement of a cementless femoral component by providing increased torsional stability for the implant and helps to maintain more normal proximal femoral anatomy25. Proximal femoral anteversion is corrected by rotating the femoral fragments, which leaves the greater trochanter and the abductors in an anatomic lateral position and restores the abductor lever arm to minimize instability and limping. In addition, subtrochanteric osteotomy makes it possible to avoid the problems related to reattachment and healing of the greater trochanter26. Nonunion rates of between 8%27 and 29%11 have been reported following greater trochanteric osteotomy with associated proximal femoral shortening performed during total hip arthroplasty in patients with congenital hip dislocation. Lastly, subtrochanteric osteotomy allows intraoperative adjustment of limb length12. The main disadvantage of the osteotomy is the risk of nonunion.
In our series, the transverse subtrochanteric osteotomy was associated with a union rate of 93%. Previous studies have demonstrated union rates of 86% to 100% at the sites of subtrochanteric femoral osteotomies performed at the time of total hip arthroplasty in patients with dysplasia28,29. Reikeraas et al.15 attributed nonunion at osteotomy sites to implants that did not provide inherent rotational stability and were not augmented with struts, factors that may have contributed to one nonunion in our series. Yasgur et al.24 noted the importance of the rotational stability of the construct and advocated the use of either a cemented stem or a diaphyseal locking press-fit stem supplemented with struts. We also believe that rotational stability is critical, and it has been demonstrated that it can be achieved reliably with a diaphyseal locking press-fit stem (with either distal extensive porous coating or sharp anti-rotation flutes).
It is important to note that complications were not rare in this series. The most common was a nondisplaced intraoperative fracture of the distal femoral segment during insertion of the femoral component. Intraoperative fracture rates ranging from 6% to 14% have been reported in other series17,30. These fractures occurred in our series because we sought to achieve a tight press-fit of the prosthesis in the distal fragment to gain good rotational stability of the implant and the osteotomy site23. The five fractures all occurred with the use of extensively porous-coated, monolithic stems with a cylindrical distal geometry. The fractures were all treated with intraoperative placement of a wire or cable to prevent their propagation. All fractures healed and no stem became loose in the patients who had a fracture. We now routinely place a cerclage cable prophylactically before implant impaction. We continue to use both extensively porous-coated stems and modular stems with sharp distal flutes for these reconstructions, with the implant choice based on the details of the patient's specific osseous anatomy, as identified with preoperative templating. Other complications of note included dislocation in four hips, aseptic loosening of the femoral stem not associated with nonunion of the osteotomy site in one hip, and aseptic loosening of the acetabular component in one hip.
We note the following limitations of the current study. First, it was a retrospective evaluation of prospectively followed patients. Second, the study group was relatively small. However, even at our referral institution, this type of surgery represents <0.3% of all total hip arthroplasties; hence, ours was a large series of patients with this uncommon diagnosis. Third, the implants and aspects of the surgical technique in this series were not identical for all of the arthroplasties. Finally, the follow-up period is still short to mid-term, and the durability of these complex reconstructions must continue to be assessed in long-term follow-up studies.
In conclusion, cementless total hip arthroplasty with placement of the cup at the level of the anatomic hip center combined with a subtrochanteric femoral shortening osteotomy in hips with Crowe type-IV high developmental dislocation led to a high rate of stable cementless implant fixation, osteotomy site healing, and restoration of more normal limb lengths. Because of the complexity of the surgery for patients with Crowe type-IV dysplasia, the complication rate was, not surprisingly, higher than that found following total hip arthroplasty in patients with degenerative arthritis of the hip with more normal anatomy.