A Surgical Approach Algorithm for Transverse Posterior Wall Fractures Aids in Reduction Quality

Abstract

Background

Transverse posterior wall fractures are difficult to treat and historically have been associated with stiffness, posttraumatic arthritis, and pain, which correlate with the reduction. The Kocher–Langenbeck approach is used most often, whereas the extended iliofemoral approach has been reserved for more complex injury patterns. The latter approach has substantially more risks. No data to our knowledge exist on the use of sequential anterior and posterior approaches for this pattern.

Questions/purposes

The purpose of this study is to evaluate an algorithmic method to determine the choice of surgical approach(es) for transverse posterior wall fractures. The main question is: will this approach-based algorithm allow for adequate reduction and stabilization to union? Our secondary endpoints were Merle d’Aubigne scores, reoperations, and radiographic sequelae including arthritis, avascular necrosis, and heterotopic ossification.

Methods

A retrospective study was conducted in which patients were drawn from an existing database. The inclusion criterion was transverse posterior wall fractures with adequate imaging treated by one surgeon. All but one patient were treated within 2 weeks of injury. Mean followup was 23 months (range, 3 months to 11 years). Between November 5, 1999, and August 22, 2012, 74 patients were treated with open reduction internal fixation for this injury; nine were excluded as a result of percutaneous treatment or inadequate preoperative imaging. The remaining 65 patients (88%) comprised the study group. All patients were treated by the senior surgeon with an algorithm that consisted of either a Kocher–Langenbeck or sequential approach based on the location, magnitude, and direction of displacement of the ischiopubic segment. Indomethacin was prescribed to all patients for heterotopic ossification prophylaxis for a total of 6 weeks postoperatively. Based on the algorithm, 82% (53 patients) were treated with Kocher–Langenbeck and 18% (12 patients) with the sequential approach. Adequacy of reduction was measured using AP and Judet views of the pelvis; union was determined empirically by pain-free weightbearing and lack of displacement over time. Outcomes were the Merle d’Aubigne score and radiographic findings of avascular necrosis or arthrosis.

Results

The algorithm resulted in 100% reduction within 1 mm on plain radiographs. Initial displacement was greater in the patients undergoing the sequential approach (p = 0.01, 7.7 versus 12.4 mm). The average d’Aubigne score was 15.3. Radiographic arthritis scores were 68% excellent/good. Avascular necrosis developed in five patients (8%). Five patients (8%) went on to THA, and four patients (6%) developed superficial or deep infection. Only one patient developed Brooker III heterotopic ossification and this was not symptomatic.

Conclusions

This algorithm helps guide appropriate selection of the surgical approach and results in accurate reduction with functional and radiographic results that are comparable with existing series while avoiding extended approaches. However, like any operative decision, the choice of approach should not depend entirely on an algorithm; rather, the algorithm is best used as a guide to understand the factors involved in treating these rare and complex injuries and to help make an appropriate choice for an individual patient.

Level of Evidence

Level IV, case series. See the Guidelines for Authors for a complete description of levels of evidence.

Introduction

Transverse posterior wall fractures of the acetabulum have poorer results than most other patterns even in the hands of experienced surgeons. In Letournel and Judet’s [2] series of 117 patients with transverse posterior wall fractures treated operatively within 3 weeks, they used the Kocher–Langenbeck approach 77% of the time (90 patients). A perfect reduction was obtained in 79 patients, 12 of whom (15%) developed radiographic arthritis. An imperfect reduction was obtained in 23 patients and 17 (74%) developed arthritis. Their book recommends the extended iliofemoral for transtectal fractures and a Kocher–Langenbeck for juxta- and infratectal patterns. Matta [4] reported an 80% anatomic reduction rate (48 of 60 cases) for transverse posterior wall fracture types and fair or poor clinical results as measured by d’Aubigne score in 30% (18 of 60 cases). Like Letournel and Judet, he used the Kocher–Langenbeck approach (77% of cases, 46 of 60 patients) with the option for extended iliofemoral in transtectal patterns or if there was an extended posterior wall fragment. In the second largest reported series of 104 transverse-posterior wall patients, Gansslen et al. [1] used the Kocher–Langenbeck approach 90% of the time (94 cases) with 76% (79 of 104) obtaining an anatomic reduction, which is similar to Letournel and Judet’s series. The Merle d’Aubigne score is the most frequently reported outcome assessment for these fractures, and in these series, approximately one-third of patients have fair or poor results, although the scores can vary highly between series [1–3].

These data suggest an overwhelming preference for the Kocher–Langenbeck approach with a higher percentage of imperfect reductions in this pattern than in others. Not even the strategy of using the extended iliofemoral approach for treating complex patterns improves transverse posterior wall fractures’ results substantially; a 2012 study by Tannast et al. [6] listed the use of this approach as a predictor of poor clinical results despite accurate reductions. Thus, it is clear that the Kocher–Langenbeck approach is ineffective in gaining the reduction even in the best hands for certain patterns, and the extended iliofemoral is a more morbid approach with its own set of problems. Above all else, the factor most under the control of the surgeon is the reduction, which correlates with the risk of arthritis in all studies of acetabulum fractures.

The senior author (PT) has used either a sequential (ilioinguinal and Kocher–Langenbeck) or an isolated Kocher–Langenbeck approach based on an algorithm related to the location, magnitude, and direction of displacement of the ischiopubic segment. Displacement greater than 1 cm and a higher exit point through the anterior column lead to a sequential approach (Fig. 1). No extended approaches have been used. The main purpose of this study is to evaluate the use of this algorithm on the quality of the reduction achieved; specifically, we asked whether an algorithmic approach would allow for accurate reduction and stabilization to union for transverse posterior wall acetabular fractures. Secondary questions were: What are the rates of early radiographic arthritis and other complications, and what are the functional outcomes for the patients treated with such a protocol?

Fig. 1.

Algorithm for treatment of transverse/posterior wall acetabular fractures. Displacement direction (anterior/posterior/neutral) refers to the sagittal plane translation of the free fragment for the transverse component measured on an axial CT scan at the level of the acetabular roof. Obliquity is measured on AP and Judet radiographs and refers to where the transverse fracture line exits at the columns. For example, if the fracture exits high at the anterior column and low on the posterior column, it is classified as “high anterior”; if exiting at the same level, it is classified as “neutral.” *One patient out of seven was treated with the sequential approach (did not follow algorithm); the patient had sacroiliac joint displacement anteriorly; ^†one patient out of six treated with the sequential approach (did not follow algorithm); the patient had a very high posterior wall fracture and the surgeon worried about correct placement of tines from the back.

Patients and Methods

This study evaluated a single surgeon’s (PT) series of operatively treated transverse posterior wall acetabular fractures performed at one institution over a 13-year period (1999–2012). Institutional review board approval was obtained for the study. Patients were identified through the orthopaedic department’s existing database of acetabular fractures by searching for the transverse posterior wall fracture pattern. The algorithm described in this report was in place during the entire period of this study and was applied to all patients.

The choice of operative approach for open reduction and internal fixation was determined based on the predicted ability to reduce the fracture using the Kocher–Langenbeck approach in the prone position. In particular, more vertical patterns and those with greater anterior displacement were thought to be more difficult through the Kocher–Langenbeck approach and favored an anterior approach for reduction of the transverse portion followed by a Kocher–Langenbeck for open reduction and internal fixation of the posterior wall. More specifically, all juxtatectal and infratectal fractures were treated with the Kocher–Langenbeck approach, whereas the approach to transtectal fractures relied on further assessment of radiographs and CT scans (Fig. 1).

The anterior approach used was the ilioinguinal on a traction table in all cases. The lateral and middle windows were used. The anterior portion of the transverse fracture was reduced using an offset clamp from the quadrilateral surface to the external surface of the ilium to correct medialization. If this reduction maneuver caused distal displacement of the ischiopubic segment, then a 3.5-mm pelvic reduction forceps was used on each side of the iliopoas muscle to correct cranial-caudad displacement along with the offset clamp. The reduction was evaluated under direct vision and fluoroscopically. Once the reduction was felt to be anatomic, it was fixed using a percutaneously placed retroacetabular screw to the quadrilateral surface that allowed removal of the clamps. The reduction was confirmed and then supported with an anterior column plate.

Direction of displacement, as defined by the direction of translation of the ischiopubic fragment on axial CT, was the most important determinant followed by the obliquity of the fracture line (where it exited in the columns). The majority of fractures (82% [53 of 65 patients]) were treated with the posterior approach. The sequential approach (Fig. 2) was used when the fracture exited high in the anterior column, for significant displacement, and for anterior displacement that would be difficult to reduce from the back without adequate visualization. The presence of posterior wall comminution, dislocation, marginal impaction (48 patients, of whom 52% were grafted), or femoral head injury had no bearing on the algorithm. All patients were prescribed postoperative indomethacin for 6 weeks for heterotopic ossification prophylaxis. Analysis of the data shows a significant difference for amount of initial displacement between patients who were treated with a posterior versus sequential approaches (Table 1).

Fig. 2A–G.

An example of using the algorithm to treat a left transverse + posterior wall fracture in a 41-year-old man. This patient was treated with sequential approaches. (A) AP pelvis radiograph showing transtectal transverse fracture line (arrow, first step of algorithm). (B) Axial injury CT scan showing gapping but no translation in the sagittal plane (arrow). This is classified as neutral displacement in the sagittal plane (second step of algorithm). (C–D) Judet views showing fracture (arrows) exiting high in the anterior column and relatively low in the posterior column. This is classified as “high anterior obliquity” (third step of the algorithm). The large gap and high obliquity of the fracture favor a sequential approach (final step of the algorithm). (E) Intraoperative fluoroscopy showing reduction of the transverse fracture. (F) The final construct for the sequential approach is seen here. The mild symphyseal separation was not addressed because it was aligned after reduction of the transverse fracture. (G) Followup radiograph taken approximately 1 year after injury. The joint space is maintained and the patient had a score of 18 at 3 years.

Table 1.

Injury and outcome data separated by surgical approach

Variable	Total patients (n = 65)	Posterior approach (n = 53)	Sequential approach (n = 12)	Statistical analysis: posterior versus sequential
Initial injury displacement* (mm)	7.7 ± 7.667	6.6 ± 7.744	12.4 ± 5.397	p = 0.01
Reduction quality (mm)	0.13 ± 0.308	0.13 ± 0.312	0.13 ± 0.311	p = 1
Final joint space narrowing (mm)†	0.87 ± 1.26	0.9 ± 1.26	0.78 ± 1.3	p = 0.8
	(n = 41)	(n = 32)	(n = 9)
Final modified Merle d’Aubigne score‡	15.3 ± 2.452	15.1 ± 2.398	16.3 ± 2.732	p = 0.25
	(n = 41)	(n = 35)	(n = 6)
Avascular necrosis (number)	5	5	0	N/A
Heterotopic ossification (number)	12 (18%)	7 (13%)	5 (42%)	p = 0.04
Total hip arthroplasty (number)	5 (8%)	4 (8%)	1 (8%)	= 1

Values are mean ± SD or number with percent in parentheses; significance set at p < 0.05; * measurements made after the hip dislocations were reduced; ^†patients who had less than 1 year of followup and 0-mm joint space narrowing were excluded to give a more conservative estimate of radiographic arthritis; ^‡only included patients with available data, number is listed in each group; N/A = not applicable.

To be included, patients had to have adequate imaging including radiographs and a preoperative CT scan and have been treated with open reduction and internal fixation. Patients were excluded if they did not have adequate imaging or an operative note available or if they were treated nonoperatively or percutaneously. Importantly, “T”-type + posterior wall fractures were excluded to allow for a more pure discussion of the transverse posterior wall pattern because the vertical limb of T-shaped fractures may change the planned surgical approach.

During this period of time, 114 patients with transverse posterior wall fractures were identified in the database. All were examined for eligibility; 49 were excluded (22 treated nonoperatively, nine treated operatively but with inadequate followup/preoperative imaging, eight were T-shaped, six not treated at our facility, two died early in the hospital course, one treated percutaneously, and one treated by a different surgeon). In total, 74 patients were treated with open reduction and internal fixation by the senior author; after nine of these were excluded as described previously, 65 patients (88%) remained for final analysis. There were 48 men and 17 women with an average age of 36 years (range, 18–64 years). Injury Severity Score was available for 52 patients with an average score of 14. Forty-two patients (65%) had an associated posterior hip dislocation and eight (12%) had an ipsilateral long bone injury. Average time to surgery was 4 days (range, 0–23 days) and all but one patient was fixed before 2 weeks. Mean followup was 23 months (range, 3 months to 11 years). Data collected included demographics (age, sex, baseline ambulatory status), injury and surgery characteristics, complications, and revision surgery. Medical records, including admission documents, operative notes, and followup clinic notes, were obtained and data were extracted for surgical approach (either Kocher–Langenbeck or sequential), type of fixation construct, and use of bone graft. Images (radiographs and CT scans) for each patient were evaluated for fracture subtype (transtectal, n = 28; juxtatectal, n = 31; or infratectal, n = 6), hip dislocation, direction and amount of displacement after any dislocation was reduced, presence of impaction or intraarticular fragments, final reduction, and subsequent union. Anatomic reductions were studied by the senior author and defined as the fracture site not being visible or being visible on radiographs with ≤ 1-mm displacement and no joint stepoff (as per Matta [4]). Postoperative CT scans were not routine, so the posterior wall was evaluated only on the five views of the pelvis (as was the case for all other series). Complications of hardware failure, joint space narrowing, heterotopic ossification, and avascular necrosis were studied from chart review and postoperative films by the primary author. Union was determined empirically by pain-free weightbearing and lack of displacement over time. Outcomes were the Merle d’Aubigne score (as per Matta [3]; the ROM score was modified by asking the patient for percentage of the normal side if they could not return to the clinic for examination) and radiographic findings of avascular necrosis or arthrosis. For radiographic arthritis, joint space narrowing as compared with the unaffected side was recorded in millimeters. Arthritis was graded using the Matta criteria where “excellent” indicated no difference from the unaffected side, “good” indicated mild changes and small osteophytes, “fair” indicated < 50% joint space narrowing and moderate changes/sclerosis/osteophytes, and “poor” indicated advanced changes, large osteophytes, > 50% joint space narrowing, acetabular wear, or head collapse [4]. If the patient received an arthroplasty during their clinical course, this was also graded as “poor.” Of note, in the analysis of radiographic arthritis and joint space narrowing, patients with < 1 year of followup who did not have joint space narrowing (23 of 65 [35%] were excluded). Patients who had joint space narrowing before 1 year of followup were included to give a more conservative estimate of arthritis. Using a scripted questionnaire, we also made an effort to contact each patient by telephone to obtain the most updated data on functional status when they could not return to the clinic.

The data were entered into a secure spreadsheet by the investigators and analyzed by the primary author (YB). Percentages, means, and SDs of variables were calculated using the mathematical functions within Excel (Microsoft Inc, Redmond, WA, USA). Differences in outcome between the Kocher–Langenbeck and sequential approaches were analyzed. Fisher’s exact test was used for categorical variables. Student’s t-test was used for continuous variables. Statistical significance was set at p < 0.05.

Results

The algorithm used resulted in all 65 patients having a reduction with ≤ 1 mm displacement on the AP and Judet radiographs (anatomic by Matta [4]) (Table 1). The average final joint space narrowing was 0.87 mm, which, with the numbers available, was not different between the two approaches. Neither the presence of posterior wall comminution nor impaction influenced the amount of joint space narrowing with the numbers available (Table 2). When the Kocher–Langenbeck portion of the sequential approach was performed, the posterior portion of the transverse fracture was examined and in no case was there residual displacement.

Table 2.

Joint space narrowing as related to simple versus comminuted posterior wall fractures and presence versus absence of acetabular impaction^*

Variable	Joint space narrowing (mm)	T-test analysis
Simple posterior wall (n = 12)	0.38 ± 0.8	p = 0.1
Comminuted posterior wall (n = 29)	1.08 ± 1.36
Impaction (n = 31)	0.68 ± 1.03	p = 0.19
No impaction (n = 10)	1.29† ± 1.8

Values are mean ± SD; * patients with less than 1 year followup and no joint space narrowing on most recent radiographs were excluded, leaving a cohort of 41 patients; ^†of note, the presence of higher than expected mean joint space narrowing in the “no impaction” group is likely the result of a single outlier of 4.8 mm narrowing in a patient who had a poor result. If that outlier is removed from the calculation, the mean is 0.9 mm, the SD is 1.37, and the p is 0.6, which is also not statistically significant.

On evaluation of the secondary endpoints of clinical outcomes, all patients were fully ambulatory without assistive devices before injury. At final followup, Merle d’Aubigne scores averaged 15.1 for the Kocher–Langenbeck approach and 16.3 for the sequential approach, both graded as “good.” Five patients developed avascular necrosis, all in the Kocher–Langenbeck group. Five patients went on to THA, of whom three had avascular necrosis and four had posterior wall comminution. Fifty-three patients did not develop heterotopic ossification, 11 developed Brooker I or II, and one patient (2%) developed Brooker III heterotopic ossification, which was not symptomatic enough to warrant excision. Two patients developed superficial and two developed deep infections. Of the deep infections, one was a polytrauma patient who developed acute hematogenous seeding of three surgical sites resulting from sepsis in the perioperative period and had a good clinical result after irrigation and débridement and antibiotics without revision. The other developed a methicillin-sensitive Staphylococcus aureus infection 6 years postoperatively after doing well without narrowing for 4 years and went on to THA.

Discussion

The most common surgical treatment for transverse posterior wall fractures relies on the Kocher–Langenbeck approach with the extended iliofemoral being used for many transtectal fractures. The quality of reduction is a determinant of clinical outcome. The senior author developed an algorithm to treat these injuries that avoided use of the extended iliofemoral approach and its associated morbidity. The goal of the study was to evaluate this algorithm’s effectiveness in achieving reduction. The radiographic results were excellent with 100% of patients achieving anatomic reduction. Clinical outcomes appear to be comparable with prior studies of this fracture pattern (Table 3).

Table 3.

Comparison data of large studies of transverse and posterior wall fractures

Study	Letournel and Judet [2] (1993)		Gansslen et al. [1] (2013)		Matta [3] (1996)		Current study (2013)
Number of transverse-posterior wall (TR-PW) patients	117 TR-PW (treated within 3 weeks); all comers: n = 569		104 TR-PW		60 TR-PW; all comers: n = 271		65 TR-PW
Followup	N/A		43 months (approximately 3.5 years)		6 years		23 months (approximately 2 years)
Use of posterior approach	77%		90%		77%		82%
Union	100% union		Did not mention nonunion → 100% union		100% union		100% union
Anatomic reduction (≤ 1 mm)	77%		76%		80%		100%
Clinical score (Merle d’Aubigne)	Excellent or good	74%	Excellent or good	56%	Excellent or good	70%	Excellent or good	73%
	Fair	9%	Fair	28%	Fair	8%	Fair	10%
	Poor	17%	Poor	15%	Poor	22%	Poor	17%
Radiographic arthritis	29 cases of osteoarthritis in TR-PW fractures, not separated by grade		Excellent or good	70%	Excellent or good	77%†	Excellent or good	68%‡
			Fair	15%	Fair	11%†	Fair	20%‡
			Poor	15%	Poor	13%†	Poor	12%‡
Heterotopic ossification (all grades)	29%		1%		18%†		18%
Heterotopic ossification (Brooker III and IV)	14%		1%		9%†		2%
Avascular necrosis	3%		10%		3%†		8%
Infection	4%*		4%		5%†		6%
Reoperation (early)	4%* (hematoma)		7% (seroma/hematoma)		2% (partial loss of reduction)		0%
Revision	2% (THA) of n = 220 evaluated patients		10% (THA)		8% all comers (6% THA, 2% arthrodesis)		8% (THA)

* Percentage calculated for entire cohort of 569 acetabular fractures treated within 3 weeks of injury, which included 117 TR-PW fractures (variable was not broken down by fracture type in study); ^†percentage calculated for entire cohort of 271 acetabular fractures treated operatively, which included 60 TR-PW fractures (variable was not broken down by fracture type in study); ^‡patients who had < 1 year of followup and 0-mm joint space narrowing were excluded to give a more conservative estimate of radiographic arthritis; N/A = not applicable.

Limitations of this study include its retrospective nature, which specifically influenced the robustness of our findings in that only 41 of 65 (63%) patients have Merle d’Aubigne scores. As a result of the small sample size, our comparisons between the groups suffer from low power; however, the reduction quality data are complete and support our approach. Additional limitations include short followup (minimum, 3 months; mean, 23 months), which will underestimate the proportion of patients who may yet develop arthritis from this injury, and the assessment of reduction on radiographs only, which may underestimate posterior wall comminution and small inaccuracies in fracture reduction [5]. Finally, a small number of patients (nine of 74 [12%]) who were treated surgically did not have adequate records or followup for analysis; it is possible that these patients were not doing as well as those on whom we had complete followup.

Concerning the primary question of the algorithm’s effectiveness in achieving reduction and stabilization to union, our algorithmic approach allowed excellent reduction in all patients (≤ 1 mm) and 54 of the 65 patients (83%) had no visible fracture line. In comparison, Letournel and Judet [2], Matta [4], and Gansslen et al. [1] reported 68%, 80%, and 76% excellent reductions, respectively. Of note, our rate of using the posterior approach alone was similar to other studies (our study 82%, Matta 77% [4], Gansslen et al. 90% [1], and Letournel and Judet 77% [2]), confirming that the posterior approach is effective in treating most of these fractures, but it requires appropriate indications. In particular, high angle transtectal fractures with greater anterior displacements are well reduced with the sequential approach.

Our results in terms of clinical scores and radiographic arthritis appear similar to earlier reports (Table 3). Our frequencies of infection, arthroplasty, and avascular necrosis also appear similar to those reported in other studies [1–4]. One notable exception is the lack of clinically significant heterotopic ossification (Brooker Grade III and IV). Our study had only one patient with Brooker III and no excisions of heterotopic ossification were performed; this is a much lower rate than in either Matta’s or Letournel and Judet’s study (the Gansslen et al. study [1] had one patient with Brooker IV, and it was not stated whether this patient received excision). This may be the most important finding, because the extended iliofemoral approach used in other studies is historically associated with a high proportion of patients developing heterotopic ossification (57% before radiation in Letournel and Judet’s series [2]). All our patients were prescribed indomethacin for 6 weeks for prophylaxis.

In conclusion, our algorithm attempts to account for the factors that make reduction difficult through the Kocher–Langenbeck approach and in those cases calls for the sequential approach instead of the extended iliofemoral. Letournel and Judet recommend the posterior approach for juxtatectal and infratectal fractures and the extended iliofemoral approach if the fracture is transtectal [2]. Matta recommends the posterior approach except in cases in which the surgeon expects difficulties with reduction such as transtectal fractures, extended posterior wall fracture, T-type/posterior wall fractures, associated symphyseal dislocation, or contralateral ramus fractures [4]. In our series, the surgeon chose a sequential method of reducing the transverse from the front and then the posterior wall from the back with a confirmation of the transverse reduction instead of an extended iliofemoral approach. The appropriate selection of surgical approach, as evidenced by our data, results in accurate reduction with functional and radiographic results that are comparable with existing series while avoiding extended approaches. However, like any operative decision, the choice of approach should not depend entirely on an algorithm; rather, the algorithm is best used as a guide to understand the factors involved in treating these rare and complex injuries and to help make an appropriate choice for an individual patient.