Introduction

Injury to the anal sphincter complex (ASC) during childbirth is a common complication and is now recognized as causing potentially long-lasting morbidity [1, 2]. Fecal incontinence (FI), a devastating and embarrassing disorder, often follows sphincter injury related to childbirth [3]. While ultrasound (US) is a vital tool in the diagnosis of anal sphincter pathology [4] 29 sonograms on each postpartum woman without signs of sphincter injury would be necessary in order to diagnosis and prevent one case of severe FI [5].

Emerging US technologies have broadened options for pelvic floor imaging and may better detect ASC pathology after the birth of a child, but this technology also displays the wide range of variation in normal sphincter anatomy. Past studies measured ASC anatomy in small cohorts of patients [6–8], but the heterogeneity of imaging methods and modalities (transvaginal, transperineal, endoanal) makes the literature difficult to meaningfully analyze and is a barrier to establishing normal anatomic ASC dimensions in parous women [4]. It is expected that measurements of normal ASC anatomy would vary based on US modality used [4, 6], but there is little information on the direction and magnitude of bias between modalities. Past studies suggest that endoanal US (EAUS) is the gold standard for evaluating the ASC compared with MRI or surgical findings [9–11]. However, recent studies indicate that the ASC can also be reproducibly evaluated with 3D translabial ultrasound (TLUS) [11–16].

Establishing normal US ASC measurements for evaluation should include both TLUS and EAUS imaging using 2D and 3D modalities. We previously published ASC measurements on TLUS in a large cohort of women 6 months after delivery of their first child by Cesarean delivery (CD) or vaginal birth (VB) [17]. In this study, we sought to compare our TLUS findings with EAUS and to compare measurements obtained with 2D and 3D imaging. We hypothesized that normal measurements would vary based on the modality used for imaging and that different muscle locations may introduce different measurement bias based on the modality being used.

Methods

This study is a planned secondary assessment of data collected as part of a parent study evaluating postpartum pelvic floor changes following the delivery of a first child. Nulliparous, healthy women who presented to prenatal care with midwives of the University of New Mexico were recruited prenatally and an additional cohort who delivered their first child by CD without entering the second stage of labor were recruited immediately postpartum. This study was approved by the University of New Mexico Health Sciences Center Internal Review Board (IRB). Informed written consent was given by all participants.

Methods are described in prior publications [13, 17, 18]. Labor and maternal characteristics were gathered at birth and included detailed examination of all lacerations to the genital tract sustained during delivery. Women with a second-degree or greater severe perineal laceration were evaluated by a second examiner to ensure that lacerations were graded appropriately. All third- and fourth-degree lacerations were repaired at the time of delivery, with standard repair methodology including identification and repair of the internal anal sphincter (IAS) with polygalactin sutures and repair of the external anal sphincter (EAS) in an end-to-end or overlapping fashion with polydioxanone (PDS) suture. All participants were scheduled to undergo 6-month postpartum 2D and 3D TLUS and EAUS imaging of the ASC. Exams were performed with the patient in the lithotomy position by one of the principal investigators (RH) or a physician trained in Female Pelvic Medicine and Reconstructive Surgery. All examiners were masked to patients’ delivery history.

Our TLUS methodology is previously described [13, 17], and displays high interrater reliability [13]. For both TLUS and EAUS imaging, the ASC was imaged in multiple planes, including proximal, mid, and distal levels [19] (Fig. 1). The proximal level was the IAS plane just distal to the anal angle; the midlevel was the location at which the pubovisceralis muscle (PVM) could be seen at its thickest point passing posteriorly to the IAS; the distal level was the level at which the IAS and EAS were optimally imaged together. For TLUS, we acquired all 2D and 3D measurements and 3D volume sets using the GE E8 US system with the 5- to 9-MHz endovaginal transducer (Milwaukee, WI, USA) or the Phillips IU22 with the 4- to 8-MHz endovaginal transducer (Bothell, WA, USA). The endovaginal transducer was placed at the posterior introitus in a transverse plane, angling nearly perpendicularly to the floor, and the ASC was imaged in multiple planes by changing the angle to image from superior to inferior. For EAUS imaging, we acquired all 2D and 3D measurements and 3D volume sets using the BK Medical Hawk system with the 10- to 16-MHz anorectal transducer (Peabody, MA, USA). The transducer was placed in the rectum, and 2D and 3D volume sets were obtained and stored in a Hawk hard drive. Each 2D and 3D volume set was stored in our imaging center’s Picture Archiving and Communication System (PACS).

Fig. 1
figure 1

Transverse (above) and sagittal (below)) images with center reference point at each of the three levels of the anal sphincter complex (ASC): proximal, mid, and distal

Full size image

In both TLUS and EAUS, we performed a full survey of the IAS and EAS complex, including imaging of the PVM (Figs. 2 and 3). The 2D TLUS transducer was rotated 90° from the axial plane to a midsagittal plane to acquire 3D volume sets. A 75° volume sweep was performed with high-quality resolution (slow sweep). We describe 3D planes as relative to the anatomy being imaged, as is consistent with standard practice in diagnostic US imaging, and is described in this manner by other authors measuring the ASC [6]. Three-dimensional volume sets were acquired in the midline sagittal plane (A plane), and 3D ASC measurements were performed in the plane axial to the midline sagittal plane (B plane). The coronal (C plane) and B planes were used to manipulate the center reference point to correctly align the overall volume. The 3D volume set was then manipulated by X, Y, and Z axis rotations in order to optimize planes to provide optimal measurements of the IAS/EAS complex at 12, 3, 6, and 9 o’clock positions in the transverse plane at proximal, mid, and distal ASC levels. We produced a subset of short field of view (FOV) 2D and manipulated 3D planes in order to optimize the thickest transverse PVM cut, most optimally measured at the pivotal 4 and 8 o’clock locations at the mid-IAS level [6, 15]. As noted in our previous publications [13, 17], this protocol did not include a FOV that was meant to optimally image PVM insertion into the pubic bone.

Fig. 2
figure 2

Endoanal (EAUS) (left) and translabial (TLUS) (right) transverse ultrasound (US) images of the distal anal sphincter complex (ASC), with the external (EAS) and internal (IAS) anal sphincters indicated by arrows

Full size image
Fig. 3
figure 3

Endoanal (EAUS) (left) and translabial (TLUS) (right) transverse ultrasound (US) images of the midlevel anal sphincter complex (ASC), with the internal anal sphincter (IAS) and pubovisceralis muscle (PVM) indicated by arrows

Full size image

We defined sphincter interruption as complete discontinuity in the sphincter at a specific location. Women with third- or fourth-degree lacerations at the time of delivery or complete IAS or EAS interruption on 6-month postpartum TLUS were included in the comparison of measurements between EAUS and TLUS imaging. Continuous variables were compared using t tests. Categorical variables were compared with chi-square or Fisher’s exact test analysis. The mean of four sphincter quadrants at each level for IAS and EAS was calculated and means compared between TLUS and EAUS using paired t tests; 2D versus 3D measurements and the left versus the right side of study participants on sphincter and PVM measurements were also compared using paired t tests. All statistical analysis was performed using SAS programming.

Results

Between July 2006 and December 2011, 782 women consented prenatally and post-Cesarean delivery to participate, and 696 delivered at the University of New Mexico hospital: 448 by VB and 246 by CD. Two women were deemed ineligible after recruitment due to having preterm vaginal births at <37 weeks gestation. There was a single forceps delivery, 25 vacuum deliveries (6 % of VB), and eight episiotomies (2 % of VB) in this study population. Ten percent of the CD patients were midwife patients transferred to physician care for CD; the rest of the CD group never entered the second stage of labor [17].

Sixty-two percent (433/694) of eligible women presented for US imaging (299 VB and 134 CD) 6 months postpartum, and 423 of them had full imaging of the anal sphincters on TLUS (299 VB and 124 CD). Characteristics of this population presenting for US, including age, birth weight, and length of the second stage of labor, have been previously reported; mean age was 25 ± 5.6 years and a racial distribution primarily of Caucasian (n = 187, 43 %) and Hispanic (n = 186, 43 %) [17]. Women who presented for US had more years of education (13.99 ± 2.74 vs 13.53 ± 2.86 years, p = 0.04) and were more likely to have undergone a VB (69 vs 56 % VB, p < 0.01) than women who did not matriculate for US imaging but were otherwise similar in baseline characteristics. Of women who presented for US imaging, 64 (36 VB and 28 CD) had full EAUS and TLUS sphincter measurements. Normative TLUS measurements of the ASC and comparison of measurements between VB and CD women are discussed in a prior publication on this cohort [17]. The majority of women who presented for US declined endoanal imaging. Women who had full US measurements of the sphincters on both EAUS and TLUS were less likely to have had a VB (56 % vs 73 %, p = 0.01),but were otherwise similar in demographics to women who had TLUS measurements only (Table 1).

Table 1 Characteristics of women presenting for ultrasound (US): women who had translabial ultrasound (TLUS) only versus women having both TLUS and endoanal (EAUS) imaging
Full size table

Sphincter interruptions on US imaging at 6 months postpartum were rare (4 EAS separations, 36 IAS separations in 38 women; 33 VB and 5 CD). Four women had interruptions of the IAS and one woman had interruption of the EAS seen on EAUS; for two of these women, the same muscle was found to be interrupted on translabial imaging. A single woman who underwent imaging in both modalities had discrepant readings; interruption of the EAS was seen on endoanal imaging; however, translabial imaging recognized the IAS as interrupted but did depict EAS interruption. Mean TLUS and EAUS measurements of anal sphincter thickness averaged over the four quadrants differed, although these differences were small (Table 2). The EAS was larger on EAUS than on TLUS, whereas all IAS measurements were smaller on EAUS (all p < 0.01).

Table 2 Endoanal (EAUS) versus translabial (TLUS) ultrasound measurement of the anal sphincter complex (ASC)
Full size table

There were 388 women with both 2D and 3D translabial measurements of the PVM to compare using the paired t test: 3D imaging yielded similar measurements compared with 2D imaging alone on both the right (7.06 mm vs. 6.97 mm, p = 0.24) and left (7.33 mm vs. 7.22 mm, p = 0.21). sides There were 137 women who had both EAUS and TLUS imaging of PVM thickness with no significant differences on the right (p = 0.69) or the left (p = 0.18). When comparing left to right sides for both US modes, the right side of the distal IAS was significantly thicker than the left (9 o’clock thicker than 3 o’clock) on EAUS; this was also true for IAS mid and distal levels on TLUS (Table 3). The EAS was thicker on the right side on TLUS (3.04 mm vs. 2.84 mm, p < 0.01) and on EAUS (3.78 mm vs. 3.66 mm, p < 0.01). On TLUS the left PVM was thicker than the right for both 2D and 3D measurements (p < 0.01).

Table 3 Asymmetry (Left versus Right, or 3 o’clock versus 6 o’clock) in sphincter measurements
Full size table

Discussion

We found small but significant differences in ASC measurement on EAUS versus TLUS in women 6 months after the birth of their first child; EAUS measured the EAS as larger and the IAS as smaller than TLUS measurement in this cohort. The study also demonstrates that women tend to have thicker anal sphincter measurements on the right side, regardless of whether imaging is translabial or endoanal, and that PVM measurements tend to be greater on the left. These data describe where and how clinicians can expect measurements to differ between imaging modalities. However, the measurement variation was small and unlikely to be clinically significant, so these data would support clinicians using the imaging modality best suited to their experience and to patient comfort.

EAUS has long been referred to as the reference standard for evaluating these muscles. EAUS has been found to be comparable with MRI and has advantages in imaging certain planes [7, 9, 20, 21]. Recent investigations, based on small numbers of women, demonstrated that 3D EAUS can image the pelvic floor in motion, such as during pelvic floor contraction or Valsalva, allowing evaluation of changes in the genital hiatus and organ descent [8, 10]. However, other studies found that TLUS can also be used to examine these dynamic changes, with notable reliability, and can demonstrate important findings such as postsurgical dynamic changes in the degree of pelvic organ descent [16, 22]. A recent study by Roos et al. discussed the comparison of TLUS and EAUS for IAS and EAS defect evaluation in 161 postpartum women, but their study did not determine normative measurements or compare measurements between modalities [9]. Another study described findings in the same women with both EAUS and transperineal US and were able to see defects well on both types of imaging, although no measurements were taken or compared [23].

A previous study from our institution established the measurements for TLUS measurements of the ASC in a large series, and these measurements were found to be highly reliable in a series of 60 women without pelvic floor disorders [13]. This reliability was also established for evaluating the levator musculature [24]. Data from future and ongoing studies may confirm that the ease and comfort of TLUS is accompanied by reliability and accuracy equal to that of EAUS, but this is yet to be established. In this cohort of young primiparous women, the majority of patients declined endoanal imaging, indicating it may not be acceptable to many patients. The large number of women who had TLUS but did not agree to EAUS, and the sole woman who had an EAS defect recognized on one modality but not another, precludes our commenting on which of these methods may be superior for sphincter defect recognition. However, we found that both modalities yielded similar ASC measurements, although modality direction bias depends on the muscle being imaged.

There is debate as to whether asymmetries seen in US imaging are a product of transducer location or part of normal female anatomy. This study’s findings are consistent with endoanal MRI data, demonstrating that the anterior portion of the sphincter in women may have less length [25]. It has been suggested that these findings may be due to EAUS transducer circumferential stretch of the musculature, making these measurements unreliable; however, the fact that this does not affect all quadrants equally is puzzling to investigators. Past studies have established that EAUS measurements of the IAS may also be unreliable, given the additional stretch of the IAS by the endoluminal transducer [26]. There is some evidence that external compression of the perineum on TLUS may distort normal anatomy,27 and this effect would theoretically be confined to the 12 o’clock sphincter position. TLUS measurements may also be altered by the change in angle involved in introital placement of the transducer. Our findings demonstrate that TLUS measured the IAS as significantly larger than did EAUS. The EAS did not demonstrate this difference on EAUS imaging, although this may be due to the fact that the EAS is lies farther from an endoanal transducer. In fact, our study indicates that EAUS measurements of the EAS are significantly larger than on TLUS. This provides evidence that EAUS and TLUS have some biases that are not only merely due to mechanical distortions, such as muscle stretch caused by the endoluminal transducer or pressure on the perineum from an introital transducer. However, the clinical significance of the measurement bias (<0.5 mm) may be quite small, even with a highly accurate exam.

Our data confirm other evidence indicating that even normal female ASC anatomy is often asymmetrical. One study demonstrated significantly more thinning of the IAS on the left side on EAUS imaging [28], which is consistent with our findings of thicker right-sided muscle measurements in both TLUS and EAUS. Our study also found that the left PVM was thicker than the right for both 2D and 3D imaging, which may be due to the handedness of the examiners or to the more common direction of the fetal occiput during birth. Literature clarifying the role of levator injury or avulsion in pelvic floor pathology has been brought to light in the years since completion of that study protocol. It is now well established that TLUS provide excellent images of the PVM and establish avulsion [8, 22, 29], but there may be more PVM injury or thinning on one side than the other, with more thinning on the left reported in a previous study [30]. These findings of left to right asymmetry could be due to our imaging methodology, asymmetrical sphincter, or levator ani damage in the birthing process; or it may be an unexpected part of normal female anatomy. With the ability to manipulate the 3D volume set, as in this protocol, measurement planes can be fine tuned to best visualize asymmetry only if it truly exists. However, this protocol was only intended to measure PVM thickness and did not capture all clinically relevant injuries, such as levator avulsion. It is recommended, given the asymmetry found in women from one side to the other, that imaging methodology seeking to represent the anatomy should always involve both sides using a US protocol that can fine tune the image.

This study is limited by its single-institution design and inclusion of only primiparous, low-risk women, and therefore may not be applicable to all populations. Also, most women in our study who presented for postpartum US did not allow the examiner to perform EAUS imaging, indicating that, for many women, this modality may be perceived as too uncomfortable and invasive. A strength of this study is the inclusion of a low-risk obstetric population, with relatively low rates of episiotomy, operative vaginal delivery, and severe laceration, making these measurements generalizable to many low-morbidity postpartum patients. If EAUS is not a comfortable or available option, we established that TLUS measurements will be similar to those found on EAUS, with bias of small magnitude only and direction bias depending on the muscle imaged. Providers should use the imaging modality which they are most experienced at performing and interpreting and should consider patient comfort and location of the suspected pathology when choosing an imaging modality.