All women underwent magnetic resonance imaging (MRI) of the pelvic floor using proton density T-2–weighted scans; 2-dimensional fast spin proton density MR scans were performed at 5 mm intervals in the axial, sagittal, and coronal planes in the supine position using a 1.5 Tesla superconducting magnet (Signa; General Electric Medical Systems, Milwaukee, WI). In addition, a pelvic examination to evaluate prolapse and a multichannel urodynamic evaluation were performed. The mean age of the subjects in our database was 46.4 years (SD 15.4), mean body mass index (BMI) was 26.3 kg/m² (SD 5.5), and mean parity was 1.9 (SD 1.7). Ninety-three percent were white, 4% were black, and 3% were classified as “other.” Race was determined by self-identification. In these ongoing studies, 155 women (23%) were recruited as cases of prolapse, and 154 (23%) were recruited as cases of stress urinary incontinence. The remaining 54% were asymptomatic controls.

Levator ani muscle defects in all MR scans had been scored by 2 independent investigators, using our previously described levator scoring system.⁴ The left and right pubovisceral muscles are scored separately. Each muscle is assigned a score from 0 to 3. Normal muscle is given a score of 0. Injured muscle is given a score of 1 if less than half of the muscle is missing, a score of 2 if more than half the muscle is missing, and a score of 3 if complete muscle bulk is lost. The inclusion criterion for this study was a complete unilateral defect, defined as a levator score of 0 on 1 side and 3 on the opposite side. Our database was queried to select only patients with a complete unilateral defect; 14 subjects met this criterion.

To visualize the complex three-dimensional (3D) geometry of the levator muscle, axial, sagittal, and coronal MR images were imported into a 3D imaging program (3D Slicer, version 2.1b1, Brigham and Women's Hospital, Boston, MA) and aligned using anatomic landmarks. For all 14 subjects, 3D models of the following structures were created from axial scans: pubic bone, levator ani muscle, obturator internus muscle, urethra and bladder, vagina and uterus, perineal body, rectum and internal anal sphincter, and external anal sphincter. We then used the following technique of missing muscle mapping to characterize the absent muscle. In the axial plane, normal muscle was visualized on one side and compared with the contralateral missing muscle. We traced the outline of that portion of intact muscle whose corresponding mirror image was absent (Figure 1), and created a 3D model of the missing muscle (Figure 2).

In all 14 cases of women with a complete unilateral levator defect, the damaged portion of the muscle was identified as part of the pubovisceral muscle. None had concurrent damage noted in the iliococcygeal or puborectal portions of the levator.

The overall direction of the missing muscle is diagonal from anterosuperior to posteroinferior (Figure 2). In all cases, the missing muscle originates from the posterior aspect of the pubic bone near the midline and extends laterally toward the iliococcygeal muscle. The fibers missing are those that arise from the pubis lateral to the pubic symphysis and over the anteromedial aspect of the obturator internus muscle (the arcus tendineus levator ani). This feature is best seen in Figure 2, C and D. The most anterior portion of the missing muscle is lateral to the space of Retzius and medial to the obturator internus muscle. The midportion of the muscle is lateral to the vagina, with attachment to the vaginal wall (Figure 3). The most posterior part muscle is lateral to the internal anal sphincter and rectum and medial to the ischiorectal fat.

The muscle inserts into the lateral margins of the perineal body (Figure 3). It also inserts into the intersphincteric space between the anal sphincters.

Four attachments were affected by the missing muscle in these women, including the muscular origin at the posterior pubic bone and the insertions into the lateral vaginal wall, the perineal body, and the intersphincteric space.

This study expands on the work of other investigators who have described the general location of the damaged portion of the levator ani. Hoyte et al² has used 3D models created from MR images to study the morphology and volume of the levator muscles in women with and without pelvic floor dysfunction. Using MRI, DeLancey et al⁵ identified levator defects in 20% of a primiparous population. Dietz and Lanzarone⁶ used 3D translabial ultrasound to demonstrate avulsion of the inferomedial aspect of the levator ani muscles in postpartum women. For the first time, we have demonstrated the specific anatomic connections affected by damage to the levator.

We chose to analyze only women with a unilateral levator ani defect because of the unique opportunity to study normal and abnormal muscles in the same individual. Unilateral defects themselves are not associated with any particular clinical syndrome. However, by allowing us to study the specific anatomic connections affected by damaged levator ani muscles, which have been associated with pelvic floor dysfunction,² this subset of women can help to elucidate the anatomic abnormalities that may be associated with pelvic organ prolapse.

Clinicians have noticed the gaping introitus and enlarged hiatus present in women with prolapse⁷ for many years. The origins and insertions of the missing muscle (pubis and perineal body, vagina, and anal sphincter) are directly relevant to holding the perineal body, vagina, and anus closer to the pubic bones. Loss of this muscle portion may help explain why women with prolapse, known to have a higher incidence of levator ani muscle loss,² have a larger genita hiatus.⁸

Why is this specific portion of the muscle most affected? In a computer simulation of vaginal birth, the medial portion of the pubovisceral muscle undergoes greater stretch than any other part of the levator ani muscle.⁹ Our findings of missing muscle only in the pubovisceral portion of the levator are consistent with this model. The possible connection between muscle damage and stretch at the time of vaginal birth raises the question of mechanism of injury. Whether this stretching of the muscle leads to avulsion, with the muscle tearing away from its points of attachment, or neurologic injury caused by compression or ischemic damage to the muscle and/or nerves created by direct compression of the fetal head remains undetermined.

This study provides important information about the muscle portion affected in levator ani muscle defects. However, this type of investigation has some limitations. It is not always possible to see the direction of individual muscle fascicles on MR scans. 3D model reconstruction does, however, enable us to study the morphology and the anatomic connections affected by levator damage. Using our knowledge of muscle fiber direction gleaned from the literature, anatomical dissections, and MRI studies,^3,10,11 we are able to estimate fiber direction and thereby hypothesize about the lines of action of the different parts of the levator ani muscle.

Future research will focus on the development of biomechanical models to analyze these force vectors to better understand the specific functional deficits created by damage to a particular portion of the muscle.