2D-fluoroscopic navigated percutaneous screw fixation of pelvic ring injuries - a case series

Between August 2004 and December 2007, 442 patients with pelvic injuries (pelvic ring and acetabular fractures) were treated at our department. Stable A2 and partial stable Tile B2 fractures with moderate pain at mobilization were treated conservatively. 100 of 326 pelvic ring fractures were treated operatively. In 44 patients a navigated percutaneous screw fixation of the posterior pelvic ring was performed. Inclusion criteria were all B1, B3 and C fractures, in which a closed reduction could be achieved. B2 fractures were included in cases of intolerable pain at mobilization and patient refusal of stabilization with an external fixator or in cases of contra-lateral leg and foot injuries requiring none or partial weight bearing.

Exclusion criteria for 56 patients were crescent fractures, transiliac fractures, and comminuted transforaminal sacrum fractures.

24 patients had Tile B pelvic lesions (8 × B1, 14 × B2, 2 × B3) and 20 patients had Tile C pelvic lesions (13 × C1, 2 × C2, 5 × C3). Nineteen of them were females and 25 were males (total average age 36.7 years, range 16-70 years). All data were collected prospectively.

Eleven pelvic-lesions were initially stabilized in an emergency procedure, 9 by an external fixator and 2 by a pelvic clamp. In two cases due to persistent hemodynamic instability additional pelvic packing was necessary.

Two patients were treated immediately on the day of injury according to the concept of primary total care.

The overall mean Injury Severity Score (ISS) was 21 ± 9, the Hannover Polytrauma-Schlüssel (PTS) 24 ± 7.

Twenty one patients were polytraumatized (mean ISS 27 ± 2, mean PTS31 ± 3), 18 patients were multiple injured and five had an isolated pelvic ring injury. Causes of injury included 33 traffic accidents (17 by car, 2 by truck, 7 on a motor bike, 2 on a bicycle and 5 as pedestrians), 10 falls from a significant height and 1 industrial accident with crush injury. 59% of the patients were directly admitted to our department and 41% were initially treated at other hospitals and transferred to our unit for definitive pelvic surgery.

The operation was performed under general anesthesia by two experienced pelvic surgeons, after obtaining informed consent by the patient, and giving a detailed description of the innovative nature of the method and the possibility of an intraoperative switch to an open approach. The operations were carried out as soon as possible, based on patient clinical condition.

Preoperatively, the patients received urinary-catheters for reduction of bladder volume and when possible also a bowel enema for an unobstructed visualization of the bony pelvis in the fluoroscopic images. A single dose of antibiotic prophylaxis (1.5 g Cefuroxim i.v.) was given.

Cannulated partial thread screws (diameter: 6.5-8.0 mm, length 60-130 mm, ASNIS III, Stryker, Germany) were used as lag screws. The screws were placed over a 3.2 mm navigated guide wire. In cases of symphysis diastasis, these lesions were usually addressed first by plate-fixation (Matta Pelvic System, Stryker, Germany or Low Profile Pelvic System, Synthes, Germany). Depending on the type of fracture, SI-screws were placed in the first or second sacral vertebra (S1 or S2) as well as ramus pubic screws via a retrograde approach, as previously reported [3, 11, 14, 22, 27, 28].

Patients were placed on a standard table (No.1150.30A0; Maquet, Germany) with a radiolucent carbon back plate and a plate extension (No.1150.45B0 and 1150.46B0; Maquet, Germany) in supine position with a sterile draped free moveable leg on the injured side to allow intraoperative manipulation for indirect reduction-maneuvers. External fixators or Schanz-screws were used as reduction aids (joysticks). The advantage of the supine position is the simultaneous reduction procedure of the anterior and posterior injuries. In 6 cases where a sufficient evaluation of the achieved reduction was not possible by 2D fluoroscopic images, an additional intraoperative 3D-fluoroscopic scan was performed, confirming an adequate reduction in each case (<5 millimeter dislocation). Otherwise a conversion to an open reduction would have been the next step in the protocol.

Fixation of the minimal invasive array (MIA) via an additional skin incision on the opposite side of the intact pelvic ring (anterior iliac crest) was usually performed as the first operative step. Fluoroscopic images confirming the reduction in up to four different projections (depending on screw position: a.p., true lateral, inlet, outlet, ala inlet or obturator outlet) were obtained and transferred to the navigation system. The screw-position was planned on a sterile covered touch screen using these images (Fig. 1). After calibration of instruments and verification of their position in relation to the anatomic object, the navigation procedure was started. The entry point on the skin was determined by holding the navigated drill sleeve on the skin surface in the planned screw axis. Then a 2 cm skin incision and blunt preparation to the ilium were performed. During insertion, the pathway of the 3.2 mm guide wire through the different fluoroscopic planes was controlled virtually. Additional fluoroscopic images were only obtained before reaching the plane of the ipsilateral sacral neural foramina and at the final guide wire placement to control the position. The length of screws was measured by a reverse ruler. After drilling, the screws were fitted with a washer and placed with a cannulated screw driver.

Pre- and postoperative CT scans (2.5 mm slices, Light Speed, General Electric, Germany) were performed for fracture classification and planning of the surgical procedure, as well as for postoperative evaluation of the reduction and screw placement (Fig. 2) [29]. The screw position was assessed using three categories:

I. secure placement, completely in the cancellous bone (Fig. 2a)

II. secure placement, but contacting the cortical bone (anterior sacrum, neural foramina or spinal canal) (Fig.2b)

III. malplacement, penetrating the cortical bone (Fig.2c).

A 3.2 mm guiding sleeve was navigated based on 2D-fluoroscopic images (Ziehm Vario, Ziehm Imaging Inc., USA) using a passive optoelectronic navigation system (Vector Vision, Traumasoftware version 2.5.0 and version 2.6.1 since January 2007, Brainlab, Germany) (Fig.1). The technique of this procedure was described in detail by Kahler et al. and Stöckle et al. [11, 15].

Physiotherapy was started on the second postoperative day. Depending on the type of fracture, the patients were mobilized on crutches. Pain orientated full weight bearing was allowed immediately for Tile type B fractures, partial weight bearing (15 kg) for 6 weeks for all Tile type C fractures, except C3 fractures treated by bed rest for 3 weeks. Postoperatively, patients were seen for a physical examination at 6 weeks, 3, 6 and 12 months post-surgery and annually thereafter. At 6 weeks and 1 year follow up, radiological examinations and additional questionnaire evaluations using the SF-36 were administered (Tab.1) [30].

Results are expressed as means ± standard error. Differences were analyzed by a simultaneous comparison step-down-procedure of Bonferroni und Holm. P-values less than 0.05 were considered as significant. The study and follow up examinations were approved by the institutional ethics committee (No.2131-10/07).

The first report of an open screw fixation of a sacroiliac dislocation was by Lehmann et al. (1934) [31]. Since then, the technique has undergone several modifications, based on Matta's widely established and standardized fluoroscopic projections (inlet and outlet) to evaluate the pelvic ring lesion and control the screw placement intraoperatively [2, 14, 32]. An additional true lateral view further increased the accuracy of transiliacal screw placements, especially for the screw placement in S2 [33].

Due to the limitations of 2D-fluoroscopic images, a repetitive change of the c-arm projection (90° to another) is mandatory to visualize the guide wire position in all three dimensions during drilling.

If a correction in one plane is necessary, a deviation in the third unobserved dimension can occur. In these situations, the guide wire has to be retracted and repositioned.

This procedure was described as "trial and error placing" by Kahler et al., resulting in high radiation exposure for the patient and the surgeon [11]. Even the most experienced surgeons reported mean fluoroscopy times of 126 seconds for placing iliosacral and 200 seconds for placing retrograde superior pubic ramus screws [4, 17]. In some cases a mean fluoroscopy time of up to 18 minutes was needed for iliosacral and transiliacal screw positioning [22].

To avoid time-consuming changes of the c-arm position, the simultaneous use of two fluoroscopes was reported. This significantly reduces the fluoroscopy time (320 vs. 270 seconds/screw), as well as operating time [7]. Disadvantages of this procedure are the need for two fluoroscopic units and limited space around the operating-table.

Using a 2D based navigation system, three fluoroscopic images have to be acquired once and then the movement of the guide wire could be visualized and virtually controlled on these images simultaneously. This technique reduces the fluoroscopy time and enables "first pass" guide wire placement reported in experimental settings so far [20, 26]. In our clinical application we measured a fluoroscopic time of 123 ± 12 seconds per screw. Disadvantages of this technique are the virtual display of instrument position, which can differ with the real situation in cases of unnoted reference base movement (reference frame shift). Additionally a navigation independent intraosseous guide-wire bending can occur in a depth of more than 7.7 centimeters [34]. Therefore we recommend additional fluoroscopic images before reaching the neuroforamina, as well as in the final guide-wire position, before screw placement.

A clinical comparison of operation time for percutaneous screw insertion in different studies is difficult, due to the diverse nature and type of pelvic ring lesions (non- vs. displaced anterior or posterior pelvic ring fractures, post partum pelvic ring instability) and consequently different operating algorithms [11, 23, 25, 35]. Furthermore the experience of the surgeon in pelvic surgery and in navigation procedures is an important factor, which may further influence these parameters.

As the experience of our team increased, an early definitive screw fixation was performed on the day of injury in two cases, instead of an initial fixation by a pelvic clamp or external fixator. The merits of early definitive operative treatment of pelvic fractures were recently discussed by Bircher et al. [36]. Especially closed reduction maneuvers should be performed as soon as possible to achieve optimal results.

In spite of a correct placement of screws in the intraoperative fluoroscopic control images, 6% of screw malplacements (perforation of the cortical bone) was seen in postoperative CT scans. This highlights the need for postoperative CT scans as a diagnostic tool to evaluate the screw placement [27, 37, 38]. In the literature malplacement rates for iliosacral screws are reported between 0 and 15% for the S1-position and the rate is even higher for the S2-position [15, 20, 23, 26, 35, 37, 39, 40]. Besides the known anatomic risk factors:

I sacral dismorphism [33, 41]

II malreduction [3, 42] and

III iliosacral screws placed in S2 [17, 24, 37],

one additional risk factor was found in this study:

IV bilateral iliosacral screw fixation.

In these cases the exact placement of the contralateral screw is difficult, because of the potential guide wire deviation by the first inserted screw. A higher risk of screw misplacement or selecting shorter screws has to be considered.

Nevertheless, no iatrogenic operation-related damage of neurovascular structures were seen in this study, although serious complications such as violation of the superior gluteal artery and nerve roots have been reported[18, 38, 43, 44].

Whether a 3D-fluoroscopic control scan should be recommended for intraoperative evaluation of definitive screw position, is still under discussion. Especially in one of our two cases needing a revision surgery after screw malplacement, two such 3D-scans would be required for complete imaging of the posterior pelvic ring, due to the limited field of view (maximum of 12 × 12 cm per 3D-fluoroscopic scan) [34]. Alternatively, intraoperative CT may solve this limitation, but due to high costs and demanding logistic requirements, this technology is available in a small number of trauma centers, only.

The use of recently introduced 3D fluoroscopy for navigation seems to further increase the accuracy of iliosacral screw placement compared to 2D fluoroscopy, as reported in two cadaver studies [20, 26]. The CT like visualization of osseous structures and additional reformations, displaying orthogonal planes in front of the driller tip and offering a "view ahead", seems to be an advantage, but randomized clinical studies are warranted.

Evaluating the screw loosening of the anterior part of pelvic ring, both patients were female and the fracture line was in zone I of the Nakatani system[45]. According to these known risk factors for fixation failure described by the authors, we found an obvious insufficient length of the retrograde placed screws. After modification of our treatment standards, in which the thread part of the superior ramus pubic screw has to pass the iliopubic eminence completely, no further loosening of the last screws was seen.

An overall revision rate of 9% in this study is equivalent to the study of Schep et al., who treated 12 patients with postpartum pelvic pain by using a fluoroscopic navigation system for SI-screw placement [25].

However the use of this technique is limited to specialized trauma centers so far, because of the high costs and a long learning curve, as well as requirement of experience in standard open pelvic surgery.