Evaluation of Pelvic Floor Muscle Morphology Using MR Defecography

Objectives: Weakening of the paravaginal attachments (level II) has been previously hypothesized to lead to the development of bladder descent. Avulsion of the pelvic floor muscles from the arcus tendinous levator ani (ATLA) has been specifically implicated.
Prior work in our laboratory using principal component analyses (PCA) of shape have been extremely effective in elucidating shape changes in anatomical geometries, including the pelvic floor. This study aimed to analyze pelvic floor shapes of clinically and non-clinically relevant bladder descent subjects to detect specific geometric differences that may be related to prolapse development.

Methods: We retrospectively searched the Northshore University HealthSystems’ database for patients who were diagnosed with cystocele based on MR defecography. Subjects with static axial MRIs and dynamic defecography volumes were included. The dynamic image series were used to discretize these subjects into two groups based on bladder decent (within 1 cm or more of the genital hiatus was clinically relevant).
3D-Slicer was used to segment and reconstruct the 3D geometries of the pelvic floor muscles. We then smoothed, aligned, and scaled the geometries based on corresponding points applied using Deformetrica. The PCA was performed on those corresponding points to assess geometric variability. The PCA decreases the dimensionality of each shape into a set of vectors called principal components (PC). PCs which explain more shape variation than random noise (determined via parallel analysis) were considered significant PCs. Each subject geometry is described by a value along each of the PC vectors, called a PC score, which were then used in statistical tests.

Results: This study analyzed 67 subjects who were diagnosed with radiologic cystocele via MR defecography, 28 of which had clinically relevant bladder descent. Table 1 summarizes the demographic data for both groups.
Twelve modes of variation explained more variation than random noise. After correcting for multiple comparisons, only PC 1 was significantly different between the two groups (p=0.009). PC 1 showed significant descent of the levator ani musculature (Figure 1). Specifically, the iliococcygeus muscle descended away from the ATLA. This change coincided with a noticeable increase in levator bowl volume (anterosuperior). Because these differences exist on the same PC, it implies a dependent relationship between these findings.

Table 1: Table describing the demographic and measured variables in this study. Statistically significant variables are bolded.

Figure 1: Mean shape plus/minus three standard deviations of PC 1. IC = iliococcygeus, PVM = pubovisceral muscle, PB = perineal body.

Conclusions: This study showed that subjects with clinically relevant cystocele had increased descent of the levators from the ATLA and a larger anterior pelvic area than their non-clinically relevant counterparts. These support the connection of the levator ani to the ATLA (i.e. level II anatomy) as a potentially important anatomical defect in prolapse development.