To avoid potential confounding effects of concomitant levator ani defects, the groups were also matched for levator defect status whereby half of the patients in each group had levator defects and half had no levator defects. Therefore, each of the 4 subgroups had 10 patients (Table 1).

All patients had a pelvic MR scan performed in the supine position. Multiplanar, two-dimensional, fast spin, proton density MR images (echo time 15 ms, repetition time 4,000 ms) were obtained using a 1.5 T superconducting magnet (General Electric Signa Horizon LX GE Medical System, Milwaukee, WI) with version 9.1 software. The field of view was 16×16 cm with slice thickness of 4 mm and a 1 mm gap between slices.

Axial MR images (Fig. 1) were imported into a 3-D imaging program (3-D Slicer, version 2.1b1, Brigham and Women's Hospital, Boston, MA). 3-D volume-rendering models were generated from axial images by tracing the main body of the external anal sphincter. Because the subcutaneous division of the external anal sphincter is not thought to have as much of a role in anal continence, it was not included. To perform measurements, the 3-D models were imported into I-DEAS® version 9.0 (UGS, Plano, TX), an engineering graphics program (Fig. 2). For each external anal sphincter model, a best-fit ellipse was created. Each ellipse was segmented at 30° intervals, creating 12 intersecting points similar to the face of a clock with 12 o'clock at the ventral portion of the sphincter. The cross-sectional area perpendicular to the fitted ellipse was measured at the intersecting points. Means ± standard error of the mean (SEM) of the cross-sectional areas were obtained for each of the four patient subgroups (Table 1). In the supine position, the external anal sphincter has a folded appearance. After preliminary review, it was found that folding of the sphincter produced irregular areas at 11 and 1 o'clock and so these were excluded from analysis due to concerns about the accuracy of the measurements at these locations (see Fig. 2 and the “Discussion” section). In addition, the cross-sectional area at region 6 o'clock where the anococcygeal ligament connects the sphincter to the coccyx could not be evaluated. Analysis was therefore performed by dividing the external anal sphincter into four quadrants: anterior, left, right, and posterior to detect regional variations. Ratios of cross-sectional area means were calculated to characterize the magnitude of group differences. Univariable and multivariable analyses were performed.

The mean cross-sectional areas of the subgroups are shown in Table 2. Mean cross-sectional areas did not significantly differ between women with prolapse and those with normal support regardless of levator ani defect status. However, women with prolapse and levator defects had a 49% larger mean muscle cross-sectional area compared to prolapse patients without levator defects (p=0.01). This difference associated with the defect status in prolapse patients was not seen in women with normal support. The post hoc power calculations are also shown in Table 2.

The cross-sectional area results of the four subgroups were further analyzed according to the four quadrants: anterior, posterior, left, and right (Fig. 3). In all quadrants, patients with prolapse and levator ani defects had the largest cross-sectional area, while those with normal support and defects had the smallest cross-sectional area. In patients with normal support, there were small cross-sectional area differences between those with levator defects compared to those without. However, in patients with prolapse, the difference between those with defects and without defects was much larger. This pattern was true for all locations. The larger cross-sectional area of patients with prolapse and defects was most noticeable in the anterior and posterior quadrants. In all four subgroups, the left and right quadrants had similar cross-sectional areas.

The thickness of the external anal sphincter has been measured at the 9 o'clock position using 2-D axial MR images with and without the use of an endoanal coil and found to correlate with squeeze on manometric testing [6]. External anal sphincter thickness and volume has also been quantified using 3-D endoanal ultrasonography [7, 8]. Other researchers have used 3-D models of the external anal sphincter for morphologic study and quantification. Fritsch et al. constructed 3-D models using plastinated sectional anatomical specimens of cadavers to correlate observations from histology and MRI [9]. Cornella et al. used the 3-D Slicer program to create MRI-based models of the external anal sphincters of 10 nulliparous women [10]. In these normal women, volumes of the external anal sphincter were not found to correlate with manometric pressures.

The current study extends existing research by comparing cross-sectional area measurements using 3-D models. The use of reconstructed models avoids the errors and variations of quantification using 2-D images [11]. Although volume and muscle thickness do approximate muscle function, cross-sectional area perpendicular to fiber direction is the most direct correlate of muscle force [3, 4]. Because of the higher rate of anal incontinence reported for patients with pelvic floor dysfunction such as prolapse, we would have expected to see an attenuated anal sphincter. The finding that the mean cross-sectional area was the same between women with prolapse and normal support was unexpected. The absence of sphincter defects in this study is intriguing and calls into question whether sphincter disruptions are overestimated using endoanal ultrasonography. Other authors using 3-D reconstruction techniques have also found unexpectedly low rates of sphincter disruptions [8]. Alternatively, other factors that were not measured in this study such as the internal anal sphincter function could correlate with the altered anal continence reported in women with pelvic floor dysfunction.

There is a difference between the 3-D MRI reconstructions and our preconceived notion of sphincter anatomy from endoanal ultrasonography. This is partly because with ultrasonography, the anal canal is distended by the probe and the patient is in lithotomy, while for MRI, the patients are supine with their legs together. The advantage of the supine posture results in less distortion of the external anal sphincter. The disadvantage is that the sphincter has a folded appearance, which makes analysis more challenging. While our strategy was able to account for some of the complexities of the 3-D shape, cross-sectional area measurements were limited in certain regions. For example, because the sphincter is folded, it was not possible to obtain a cross-section perpendicular at 1 or 11 o'clock. Therefore, these locations were excluded from analysis. Another limitation of this method is that the technique for model generation traces around the external anal sphincter envelope and not the muscle fascicles. The region posteriorly at 6 o'clock where the anal sphincter fascicles decussate to join the anococcygeal raphe has a feathery appearance where muscle fascicles are intermingled with fat and connective tissue. Our strategy of tracing the muscle envelope does not account for fatty infiltration of the muscle and likely overestimates the amount of muscle in this area. For this reason, the 6 o'clock location was not analyzed. In addition, the dramatically larger cross-sectional area (64%) in the anterior region of patients with prolapse and levator defects forced us to question the validity of our data because we could not explain why this quadrant would be so affected. However, the same trend of increased cross-sectional area in women with prolapse and levator defects is present in the other quadrants. This supports the validity of the result.

Despite these limitations, this technique does allow for cross-sectional area measurements perpendicular to the muscle to be obtained on the external anal sphincters of living women. This approximation of muscle function found that women with prolapse do not have reduced external anal sphincter cross-sectional areas compared to women with normal support, and women with levator injuries and prolapse actually have larger cross-sectional areas.