A twenty-four-year-old woman was the unrestrained driver of an automobile involved in a side-impact accident with another motor vehicle in August 1996. She had no history of medical problems and did not smoke. She weighed more than 100 pounds (forty-five kilograms) more than her ideal body weight. Emergency evaluation at another institution revealed a juxtatectal (directly adjacent to the weight-bearing dome) left transverse acetabular fracture with approximately five millimeters of displacement (Fig. 1). On the basis of the patient's body habitus, nonoperative treatment was recommended and the patient was not placed in skeletal traction. She was rapidly advanced to weight-bearing with a walker as tolerated, and she progressed to walking with a cane in two months. She continued to have pain and to walk with a limp. A computed tomography scan performed in January 1997 demonstrated a slightly comminuted juxtatectal transverse acetabular fracture with no evidence of healing across the major fracture line. Although her tolerance of weight-bearing improved gradually, the patient required a cane while walking outside of her home, which she did three or four days per week. Several physicians had recommended total hip arthroplasty on the basis of the persistent symptoms. She was referred to us by her attorney for another opinion approximately one year after the injury.
When first seen by us, the patient reported pain, weight gain due to inactivity, and a persistent limp. She expressed a desire to avoid both total hip arthroplasty and arthrodesis. The patient received a hip score of 61 points on the self-administered questionnaire of The Hospital for Special Surgery3. On physical examination, she demonstrated pain at the extremes of passive motion without loss of passive motion. She had a left abductor lurch when walking without a cane.
Radiographs confirmed a nonunion of the transverse acetabular fracture with preservation of the joint space and no medial subluxation of the femoral head (Fig. 2). Because of the patient's body habitus, it was anticipated that the operative exposures required for open fixation and bone-grafting of both columns of the acetabulum would be associated with a high risk of wound complications. It was recommended that computer-assisted percutaneous internal fixation be attempted. Informed consent was obtained after a discussion of all other options, including internal fixation and bone-grafting through extensile or dual operative approaches.
In September 1997, the patient was taken to the operating room for two procedures. During the first procedure, a two-pin external fixator that was fitted with four aluminum spheres (fiducials) was applied to the left ilium (Fig. 3). General anesthesia was used because the patient's body habitus prohibited placement of the frame with local anesthesia. After the patient had recovered from the initial anesthesia, she was taken to the computed tomography suite for scans of the pelvis with the attached frame (Fig. 4). The patient was then returned to the operating room for definitive internal fixation after receiving a second general anesthetic.
Before the second induction of anesthesia, a three-dimensional virtual model of the pelvis was constructed on an integrated image-guided surgical system (Stealth Station; Sofamor-Danek, Memphis, Tennessee) from the data obtained from the computed tomography. This system allows the surgeon to define a precise preoperative plan for insertion of the internal fixation pins using a line defined by identifying entry and exit points for each screw on the virtual model. The surgeon interactively modifies these planned points until the path of the screw traverses a safe anatomical zone across the fracture line. The path of intended fixation can be viewed in three dimensions on the computer workstation. The virtual path of the planned fixation screw may then be displayed in two two-dimensional surgical navigation views orthogonal to the surgical plan. These views are similar to those obtained during custom two-dimensional reconstructions of complex fracture patterns, but they are oriented along the intended path of screw placement. The length of the screw path is displayed with 0.1-millimeter accuracy (Fig. 5).
Three different surgical plans were developed for the passage of guide-wires for three 7.3-millimeter cannulated screws (Synthes USA, Paoli, Pennsylvania). A localizing camera interfaced with the computer guidance system looked down onto the operating field and precisely tracked the position of special surgical instruments fitted with light-emitting diodes. After a second induction of general anesthesia and sterile preparation and draping of the operative sites, an optically tracked dynamic reference array was attached to the external fixator to allow real-time tracking of the pelvis during the operative procedure. The pelvis was registered (localized) with use of an optically tracked probe that mated with each of the aluminum spheres on the external fixator. The accuracy of registration with this system is routinely less than 0.5 millimeter. An optically tracked drill-guide was then used to pass the guide-wires according to the preoperative surgical plans. The integrated computer workstation provides virtual real-time feedback, demonstrating the position of the drill-guide relative to the previously determined surgical plan. Although limited intraoperative fluoroscopy is used for validation of the registration, the need for intraoperative fluoroscopic guidance is essentially eliminated.
The first guide-wire was passed perpendicular to the major fracture line in the supra-acetabular region. Slight compression was applied with a 7.3-millimeter titanium cannulated lag screw. A second guide-wire and lag screw were passed from the posterior aspect of the ilium across the fracture into the anterior column and the superior pubic ramus in an antegrade fashion. Finally, the hip was flexed and a percutaneous retrograde posterior-column screw was passed from the ischial tuberosity into the ilium. As the pelvis was continually tracked by the dynamic reference frame, the need for repeat registration was eliminated. The external fixation (fiducial) frame was then removed. The total operative time needed for positioning, preparation, registration, placement of the guide-wires according to the surgical plan, and placement of the lag screws was 105 minutes. The total time for fluoroscopic validation of registration and confirmation of guide-wire position was thirty-five seconds. Blood loss was negligible, and the five stab incisions were closed with single sutures. Postoperative radiographs confirmed accurate extra-articular placement of the screws in accordance with the surgical plan (Fig. 6). The patient had no complications and was discharged on the second postoperative day. Only toe-touch weight-bearing was allowed for the first six weeks, after which weight-bearing was gradually increased as tolerated.
At the time of the three-month follow-up, the patient had progressed to walking without aids, the pain had greatly decreased, and the limp had resolved. At the time of the seven-month follow-up, union in the periacetabular region was noted on radiographs, although a faint lucency remained in the region of the ischial spine (Fig. 7). The passive range of motion of the hip was painless and unrestricted, and walking was unlimited. The maximum level of pain in the hip in the preceding month was described as mild, and the patient had not needed any over-the-counter or prescription medication for pain. The self-administered questionnaire of The Hospital for Special Surgery revealed a hip score of 92 points. The patient was essentially asymptomatic at the time of a fifteen-month telephone follow-up.
Letournel defined a nonunion of an acetabular fracture as a fracture that has remained unhealed for longer than four months after the injury7. However, nonunion of an acetabular fracture is rare; it usually occurs when the fracture is unrecognized or untreated or if there is persistent dislocation or subluxation of the femoral head1. Nonunion occasionally occurs after operative treatment of acetabular fractures; it occurs less commonly after nonoperative treatment. Letournel reported the treatment of eleven acetabular nonunions; six of the fractures were transverse, three involved the posterior column, and two were associated transverse and posterior wall fractures7. In their textbook, Letournel and Judet reported four nonunions (0.7 percent) in a series of 569 acetabular fractures; two were both-column fractures and two were associated transverse and posterior wall fractures8. Mears et al. reported one case of nonunion in their series of ninety-seven patients with 100 acetabular fractures10.
Letournel noted that nonunion of the acetabulum is characterized by variable amounts of pain7. Nonunion is generally apparent on radiographs, with irregularity and formation of hypertrophic bone at the site of the nonunion. Fluoroscopic examination can often reveal motion at the fracture site. Spontaneous healing of an established acetabular nonunion without operative intervention has not been described, to our knowledge. In an attempt to prevent rapid deterioration of the joint, Letournel recommended operative treatment, including decortication of hypertrophic bone from the edges of the site of the nonunion, excision of tissue at the site of the nonunion, reduction, bone-grafting, and stable internal fixation.
Computer-assisted surgery is a rapidly expanding field with tremendous potential in a variety of surgical subspecialties. The first neurosurgical procedures using computer-integrated stereotactic imaging were carried out during the early 1980s6. This technology has found other applications in the fields of ophthalmology, otolaryngology, dentistry, radiotherapy, and general surgery. In orthopaedic surgery, computer-assisted surgery has recently been applied to the placement of pedicle screws in the spine5 and to the insertion of the acetabular component in total hip arthroplasty2. Computer-assisted surgery has also been combined with a robotic interface for preparation of the femoral canal in total hip arthroplasty11. In general, however, orthopaedic surgeons have been relatively slow to embrace this new technology.
Several competing integrated image-guided surgery systems are now available. Most systems employ optical tracking of surgical instruments with digital cameras that look down onto the operating field and are interfaced with the computer workstation. Computed tomography data that are specific to the patient are loaded onto the workstation with use of a magnetic tape drive. With optical tracking of both the patient and the surgical instruments, the position of the instruments relative to the patient's anatomy can be graphically represented on the workstation monitor. A precise preoperative surgical plan is stored by identifying the entry and target points for placement of the screw on the virtual model. Two-dimensional reconstructions of the intended path of the screw can then be generated to ensure safe anatomical and extra-articular positioning of the screw and for precise determination of screw length. The surgeon can intraoperatively orient a drill-guide or other instrument with respect to a previously defined surgical plan. In a cadaver trial, this technique consistently provided sufficient accuracy to allow percutaneous passage of guide-wires into the anterior column of the acetabulum without the use of radiographic guidance4.
In our patient, it was possible to obtain internal fixation and to promote osseous healing with a computer-assisted percutaneous technique. Because of the patient's large body habitus, we believed that the risks of formal open exposure were so great that an attempt at minimally invasive fixation was justified. The nonunion site itself was not exposed operatively but was placed under compression with a lag-screw technique. A computer-guided technique allowed this to be accomplished with minimum exposure to radiation; however, two general anesthetics were required. Radiographs showed evidence of osseous healing, and the symptoms rapidly resolved. Although limited intraoperative fluoroscopy was needed for validation of registration and confirmation of safe placement of the guide-wires, the use of computer assistance greatly reduced the need for intraoperative radiographic guidance. Computer-guided operative procedures may help to decrease the need for formal open operative exposure in the treatment of selected acetabular and pelvic fractures. It is hoped that this technology will prove useful in open acetabular procedures as well, by allowing accurate blind placement of screws for internal fixation without violating the hip joint.
For percutaneous applications, it is currently necessary to place a small external fixator before computed tomography in order to obtain sufficient registration accuracy. It is hoped that this step will be eliminated in the future with the development of noninvasive registration techniques using ultrasound or fluoroscopy. With further refinements, this technology may soon be applicable to many of the minimally invasive procedures currently being performed by orthopaedic surgeons.