The 2D radiographic measurements with statistically significant differences between groups were the posterior wall distance, LCEA, and extrusion index. The posterior wall distance had near-perfect intra- and inter-observer agreement. Thus, the posterior wall distance may be used to augment the diagnosis of acetabular retroversion in patients with a crossover sign. To our knowledge, posterior wall distance has not been described previously in the literature. With high intra-observer repeatability, the LCEA may also serve as a supplementary measure. Moderate inter-observer repeatability for the extrusion index suggests that it may also have diagnostic value.
There is some disagreement in the literature as to how retroversion alters coverage. We found that retroverted acetabula provided less total coverage, slight anterior over-coverage, and substantial posterior under-coverage. Using a foam model of the pelvis, Giori and Trousdale (2003)
suggested that the crossover sign and retroversion was the result of posterior deficiency alone. Using CT scans of patients with developmental dysplasia (with a subset showing signs of retroversion), Fujii et al. (2010)
found that posterior and posterosuperior coverage was reduced in retroverted acetabula, but there were no differences between retroverted and anteverted acetabula with regard to anterior coverage. Dandachli et al. (2009)
evaluated the relationship between the crossover sign and acetabular retroversion to percent coverage of the femoral head in the retroverted acetabulum. While they found similar total coverage between normal and retroverted acetabula, they noted anterior over-coverage and posterior under-coverage in the retroverted acetabula.
Previous studies have made various, simplifying assumptions when estimating coverage, which may explain discrepancies with regard to how retroversion alters coverage. For example, Dandachli et al. (2009)
assumed that femoral heads were spherical. However, it is well known that femoral head asphericity is common in acetabular retroversion (Beck et al. 2005
, Steppacher et al. 2008
). In the study by Dandachli et al. (2009)
, half of the hips with acetabular retroversion also had a cam-type deformity. The method we used made no prior assumptions regarding the geometry of the femoral head. Instead, calculations of coverage were accurately obtained from the native geometry (Anderson et al. 2005
). Thus, our approach can account for concomitant deformities of the femoral head when calculating coverage.
While we found that anterior coverage was increased in retroverted acetabula, the amount of over-coverage (~10%) was much less than the magnitude of posterior coverage loss (~40%). Retroversion was diagnosed by the presence of the crossover sign alone. Thus, our results indirectly suggest that the crossover sign may be the primary result of deficient posterior coverage rather than excessive anterior coverage. Thus, in patients with a crossover sign as the presenting abnormality, surgery aimed at removing excess anterior wall alone may not be treating the source of the morphologic abnormality. In such cases, acetabular reorientation may effectively normalize coverage in the anterior and posterior regions simultaneously. Dandachli et al. (2009)
also tended to support acetabular reorientation as a more appropriate treatment than removing part of the anterior wall alone.
Our findings suggest that the posterior wall distance, extrusion index, acetabular index, and lateral center edge angle may predict total and regional posterior coverage. Specifically, posterior wall distance and LCEA were positively correlated with coverage, while EI and AI were negatively correlated with coverage. The directions of these correlations are logical, as increased posterior wall distance and LCEA indicate greater coverage, whereas increased AI and EI describe reduced coverage.
Posterior deficiency may cause hip pain and early arthritis through increased cartilage contact pressures in the posterior region, which lacks sufficient coverage for support (Ezoe et al. 2006
). This is consistent with the work of Bardakos and Villar (2009)
, who noted that positive posterior wall signs were 1 of 2 factors associated with progression to early osteoarthritis. The posterior wall distance we describe offers a new and simple method for characterization of posterior deficiency, as it correlated with 3D measures of total/regional posterior coverage. In addition, the high intra- and inter-observer correlation coefficients suggest clinical applicability. Finally, correlations between the posterior wall distance and total and posterior coverage were higher than for EI, AI, and LCEA. Thus, the posterior wall distance may be better suited to predicting coverage than these other radiographic measures. We found correlations between the extrusion index and both total coverage and coverage in the posterior and posterolateral regions. This measurement may therefore be useful for prediction of posterior coverage in hips where the posterior wall distance or LCEA appears normal.
Others have studied the relationship between plain radiographic measurements and CT measurements for recognition of acetabular morphology (Dandachli et al. 2008
, Werner et al. 2010
). Werner et al. investigated the relationship between the crossover ratio and CT-based measurements of retroversion (Werner et al. 2010
). They found a relationship between the crossover ratio, measured on plain radiographs, and the CT-based roof edge angle (r = –0.49, p < 0.001) and equatorial edge angle (r = 0.40, p < 0.001). However, their CT-based measurements were 2D angles, and may not fully characterize acetabular coverage like the 3D measurements evaluated in this paper. Thus, interpretation of the crossover ratio in terms of 3D acetabular coverage was not addressed. We found that the crossover ratio did not correlate with femoral head coverage. The conflicting results may be a result of the 2D angles used in the study by Werner et al.
The present study has several limitations. First, the sample size was smaller than in other studies that have used CT-based measurements to assess retroversion (Ezoe et al. 2006
, Dandachli et al. 2009
, Fujii et al. 2010
). However, as mentioned above, previous studies have used simplified analyses to calculate coverage, which may limit their applicability and require greater sample size to detect smaller differences in coverage compared to our approach. An additional limitation of our study is that a comprehensive clinical history was not available for the control hips. Thus, although hips with records or evidence of previous surgery or abnormal radiographic morphology (based on radiographic criteria) were excluded, it is not known whether these hips were asymptomatic. In addition, we did not correct all radiographs to a standard pelvic tilt or rotation. While we acknowledge that specialized software is available to correct for pelvic tilt (Zheng et al. 2007
), our study was performed without it to follow the standard diagnostic procedure for patients in our clinic and for the majority of published studies. Finally, retroversion was diagnosed by the presence of the crossover sign alone. Recently, however, Perreira et al. (2011)
demonstrated that retroversion involves the acetabulum at all levels and includes the entire pelvic segment containing the acetabulum and ischial spine.
In conclusion, acetabular retroversion was associated with a slight but significant increase in anterior acetabular coverage, especially in the anterolateral region. Retroverted hips had substantially less posterior coverage, especially in the posterolateral region. Our study showed that a number of radiographic measures of acetabular morphology were correlated with femoral head coverage. These relationships might be used cautiously by other clinicians to assist in the diagnosis of retroversion and for the purposes of preoperative planning.