How well can levator ani muscle morphology on 3D pelvic floor ultrasound predict the levator ani muscle function?

Abstract

Introduction and hypothesis

The aim of our study was to assess the performance of levator ani muscle deficiency (LAD) evaluated by 3D endovaginal ultrasound (EVUS) to detect pelvic floor muscle function as assessed by digital examination.

Methods

This cross-sectional study was conducted among 77 patients referred to our urogynecology clinic for pelvic floor dysfunction symptoms. Patients underwent physical examinations including digital pelvic muscle strength assessment using the Modified Oxford scale (MOS). EVUS volumes were evaluated and levator ani muscles were scored according to a validated LAD scoring system. MOS scores were categorized as nonfunctional (scores 0–1) and functional (scores 2–5).

Results

Mean age of participants was 56 (SD± 12.5) and 71% were menopausal. Overall, 32.5% had nonfunctional muscle strength and 44.2% were classified as having significant LAD. LAD identified by ultrasound had a sensitivity of 60% (95% CI 41%–79%) for detecting nonfunctional muscle and a specificity of 63% (95% CI 50%–77%) for detecting functional muscle. Overall, LAD demonstrated fair ability to discriminate between patient with and without poor muscle function (area under the ROC curve = 0.70 (95% CI 0.58–0.83). Among patients with an LAD score of 16–18, representing almost total muscle avulsion, 70% had nonfunctional MOS scores. Whereas, in patients with normal/minimal LAD (scores of 0–4), 89.5% had functional MOS scores

Conclusions

LAD and MOS scales were moderately negatively correlated Among patients with normal morphology or the most severe muscle deficiency, LAD scores can identify the majority of patients with functional or non-functional MOS scores, respectively.

Introduction

Injury to levator ani muscle as a result of birth has been documented [1–3]. To detect such birth-related injuries to the levator ani muscles, modern imaging techniques such as magnetic resonance imaging (MRI) [4,5], transperineal ultrasonography [6,2,7] and endovaginal ultrasonography (EVUS) [8,9] have been used. The appearance of the levator ani muscle subdivisions and a scoring system for evaluation of levator ani muscle deficiency by endovaginal ultrasound has been described [10,9]. The degree of levator ani defects seen by high resolution 3D endovaginal automatic acquisition ultrasound (EVUS) has been described as “levator ani deficiency” (LAD) [9]. Unlike the terms “defect” and “avulsion” which may imply an all or none phenomenon, the term “levator ani deficiency” implies a measurable gradient of muscle loss. The scoring system for grading LAD is based on visualization of each levator ani muscle subdivisions on each side of the pelvic floor. Each muscle subdivision is scored based on its thickness and attachment to the pubic bone. This scoring system has been demonstrated to have excellent inter and intra-rater reliability [11]. Severity of LAD detected by EVUS has been positively correlated with severity of pelvic organ prolapse [9]. Severe LAD was more prevalent in women with severe anal incontinence [12] and its role as a risk factor for urinary incontinence has been evaluated. Patients with SUI either have no LA defect or have a higher prevalence of mild LA defect as compared to moderate or severe LA defects. [13]. Additionally, LAD severity is associated with the anatomical position of the levator plate within the pelvic floor [14]; but its relation to pelvic floor function is unclear. It is to be expected that the severe deficiency should have an effect on pelvic floor contraction strength.

One of the basic forms of functional assessment of the levator ani muscle is digital palpation; however qualitative assessment based on palpation may have limited repeatability [15]. Messelink et al. recommended quantifying contractions by using the Modified Oxford Scale (MOS) to classify digital pelvic muscle strength into six categories ranging from absent to strong [15].

The aim of our study was to evaluate the performance of LAD evaluated by 3D EVUS to detect pelvic floor muscle function compared to that assessed by MOS digital examination.

Methods

Approval of this study was obtained from the Institutional Review Board at our institution. This cross-sectional study includes women who were referred to our urogynecology clinic because of different pelvic floor dysfunction symptoms including pelvic organ prolapse, urinary incontinence, anal incontinence, pelvic pain, and mesh related complications during January 2013–January 2014. The study time frame was selected because it corresponded with the dates in which we started performing and documenting MOS on all our patients. Of the 149 potentially eligible patients, 77 women had complete chart with needed data during January 2013 and January 2014 and were included in this study. Patients completed standardized questionnaires including PFDI-20 and PISQ and received a standard examination; assessment of pelvic floor by 3D EVUS, and digital assessment of pelvic floor function using MOS in their initial visit. Women with a history of prior pelvic floor reconstructive surgery and CNS or peripheral neurology diseases were excluded.

Ultrasound protocol

Imaging was obtained at the time of the primary visit using the BK Medical Ultrafocus (Peabody, MA, USA) and a 2052/8838 12 MHz transducer. All ultrasound exams were performed in the office setting, with the patient in dorsal lithotomy position, with hips flexed and abducted. No preparation was required and the patient was recommended to have a comfortable volume of urine in the bladder. No rectal or vaginal contrast was used. To avoid excessive pressure on surrounding structures that might distort the anatomy, the probe was inserted into the vagina in a neutral position with no pressure on vaginal walls. EVUS volumes were digitally stored for further analysis.

EVUS volumes were evaluated blinded to patient symptoms, and muscles were scored according to LAD scoring system, which has been previously validated. The levator muscle was divided into three subgroups; the puboperinealis/puboanalis [PA], puborectalis [PR], and iliococcygeus/pubococcygeus [PV][10]. Subgroups were evaluated and scored (0=no defect, 1=minimal defect with < 50% muscle loss, 2=major defect with >50% muscle loss, 3=total absence of the muscle) on each side based on thickness and detachment from the pubic bone. Scores were categorized as mild (scores 0–10) and significant (scores 11–18) levator ani deficiency (LAD) (Figure 1).

Figure 1.

Levator ani deficiency detected by 3D EVUS, A) normal levator ani muscles B) Significant levator ani deficiency with total score of 14.

P; pubic symphysis, R; Rectum, U; urethra, Missed LA; complete avulsion of levator ani muscle on right, Thin LA; partial avulsion of iliococcygeus on left with thinning of puborectalis

MOS Digital Palpation Protocol

Examiners were blinded to ultrasound finding while performing muscle function digital assessment. Pelvic floor muscle function was assessed subjectively by digital palpation while inserting a lubricated gloved index finger approximately 4 cm into vagina [17,18]. All women were instructed to squeeze their levator ani muscles without activation of other groups of muscles, abdominal, gluteal and adductor muscles. Examiner instructed the patient to contract correct muscles. Muscle strength was graded using the six-point MOS; 0, no contraction; 1, minor muscle flicker; 2, weak muscle contraction; 3, moderate muscle contraction, 4; good and 5, strong muscle contraction against resistance by the examining finger [15,18]. A score was used for both left and right side and the lower score was used for analysis. To compare LAD performance to digital palpation in this study, we categorized the MOS as non-functional (scores 0–1) and functional (scores 2–5).

Statistical methods

Statistical analysis was performed using SAS v9.2 (SAS Institute, Cary, NC). The distribution of patient characteristics were compared between those with mild and significant LAD and between those with functional and non-functional pelvic floor muscles using chi square tests, fishers exact tests, t-tests and Wilcoxon rank sum tests, as appropriate. The Spearman correlation coefficient was calculated to assess the association between the LAD and MOS scales. The sensitivity and specificity of ultrasound-detected LAD to identify poor muscle function was assessed using MOS as the gold standard. The kappa coefficient was calculated to evaluate agreement between two tests. A receiver operating characteristic (ROC) curve was estimated to evaluate the ability of LAD to discriminate between patients with and without poor muscle function.

Results

Mean age of the 77 participants was 56 (SD± 12.5) and approximately 71% were post-menopausal. Median body mass index and parity were 28.2 (range 20.0–49.3) and 3 (range 0–6), respectively. Pelvic organ prolapse was observed in 53.5% of patients. Among those with data available for these characteristics, 70% (51/73) had a history of prior hysterectomy and 21% (14/66) reported smoking. Women with significant LAD were more likely to be older and more likely to be menopausal compared to women with mild LAD. Similarly, women with nonfunctional MOS were also more likely to be older and more likely to be menopausal compared to those with functional MOS. Patient characteristics by LAD and MOS categories are summarized in Table 1.

Table 1.

Distribution of subject characteristics by levator ani deficiency (LAD) status and pelvic muscle strength as evaluation by the Modified Oxford Scale (MOS)

		Mild LAD (n=43)		Significant LAD (n=34)	P value
	nˆ	Mean ± sd	nˆ	Mean ± sd
Age (years)	43	51.9±12.7	34	62.4±9.5	0.0001a
		Median (range)		Median (range)
Parity	25	3 (0–6)	22	3 (1–5)	0.51b
BMI	29	27.8 (20.0–49.3)	23	28.7 (21.9–44.7)	0.38b
		n (%)		n (%)
Pelvic Organ Prolapse	38	17 (44.7)	33	21 (63.6)	0.11c
Menopause	30	17 (56.7)	21	19 (90.5)	0.009c
Prior Hysterectomy	41	25 (61.0)	32	26 (81.3)	0.06c
Smoking	37	9 (24.3)	29	5 (17.2)	0.48c
		Functional MOS (n=52)		Non-functional MOS (n=25)
	nˆ	Mean ± sd	nˆ	Mean ± sd
Age (years)	52	53.6±12.6	25	62.6±9.8	0.003a
		Median (range)		Median (range)
Parity	29	3 (1–6)	18	3 (0–5)	0.77b
BMI	35	27.3 (20.0–49.3)	17	30.8 (22.5–44.7)	0.05b
		n (%)		n (%)
Pelvic Organ Prolapse	48	22 (45.8)	23	16 (69.6)	0.06c
Menopause	35	21 (60.0)	16	15 (93.8)	0.02d
Prior Hysterectomy	50	36 (72.0)	23	15 (65.2)	0.56c
Smoking	46	10 (21.7)	20	4 (20.0)	1.00d

^ˆSample size for LAD or MOS strata with data available for the listed characteristic.

^aStudent’s t-test;

^bWilcoxon rank sum test;

^cChi-square test;

^dFishers exact test

Overall, the prevalence of non-functional pelvic floor was 32.5%. A total of 44.2% were classified as having significant LAD. The LAD and MOS scales were moderately negatively correlated (Spearman correlation coefficient=−0.40, p=0.0003). A scatterplot of the association is displayed in Figure 2. LAD as defined by ultrasound had a sensitivity of 60% (95% CI 41%–79%) for detecting non-functional muscle status and a specificity of 63% (95% CI 50%–77%) for detecting functional muscle status (Table 2).

Figure 2.

Scatter plot comparing LAD and MOS measurements

Table 2.

Comparison of LAD and Modified Oxford Scale categories

	Nonfunctional MOS (Scores 0–1) n (%)	Functional MOS (Scores 2–5) n (%)	Total n (%)
Significant LAD  (Score11–18)	15 (60.0)	19 (36.5)	34 (44.2)
Normal/Mild LAD  (Score 0–10)	10 (40.0)	33 (63.5)	43 (55.8)
Total*	25 (32.5)	52 (67.5)	77 (100.0)

Simple Kappa coefficient = 0.21

^*Row percentage

The area under the ROC curve was 0.70 (95% CI 0.58–0.83) (Figure 3), indicating LAD had fair ability to distinguish between patients with and without poor muscle function. The positive predictive value (PPV) of significant LAD for predicting non-functional MOS was 44.1% (95% CI 27%–61%) and the negative predictive value (NPV) of normal/mild LAD for predicting functional MOS was 76.7% (95% CI 64%–89%). However, when examining the extremes of the LAD score distribution, 70% of women with severe LAD scores of 16–18 (i.e., 7 of the 10 patients with almost total muscle loss) were classified as having non-functional muscle status. Among patients with normal to minimal LAD scores of 0–4, 17/19 (89.5%) were classified as having functional muscle status.

Figure 3.

ROC Curve for the model

Discussion

Our study indicates that the morphology of levator ani muscle as visualized on EVUS is moderately predictive of the function of levator ani muscles. Sixty percent of women with non-functional muscle status were classified as having significant LAD. Of women with functional muscle status, 63% were classified as mild LAD. Additionally, 70% of women with severe LAD had non-functional levator muscle, which was present in only 10.5% of women with normal morphology of levator muscles on EVUS of the pelvic floor.

DeLancey et al. compared the levator ani muscle defect and function in women with and without prolapse [19]. They used MRI for visualization and grading of levator ani muscle morphology. In their study, levator function was assessed with a vaginal speculum specially designed to record the isometric force acting on anterior and posterior bills of the speculum at rest and with pelvic floor contraction. The study quantified how vaginal closure force is affected by the presence of pelvic organ prolapse and levator defect status. Women with muscle defects were unable to augment vaginal closure force as well as those with normal muscles regardless of prolapse status. Dietz et al. studied the levator avulsion and grading of pelvic floor strength [20]. The levator avulsion was diagnosed on transperineal ultrasound and muscle strength was investigated by digital palpation based on MOS. The presence of avulsion was associated with a significant reduction in the mean overall MOS grading (1.9 vs 2.78, p<0,001). Additionally, Steensma et al. found that underactive pelvic floor muscles, which was defined as absent, or weak pelvic floor contraction on ultrasonography resulting in no or minimal reduction of the levator hiatus dimensions on contraction, occurred more often in patients with an avulsion injury [21]. Levator muscle injuries were present in 53.8% with underactive pelvic floor muscle compared to 16.1% with normal pelvic floor muscle contraction. Our study findings are in agreement with other researches results. Levator ani muscle morphology is not the only predictor of muscle function but to some extent there is a relation between impaired morphology and impaired function. The predictive value of function by EVUS was described in this study.

Our study is limited by the relatively small sample size. Additionally, three clinicians performed MOS digital examination for pelvic floor function assessment; thus, inter-rater variability could have introduced some error in the assessment of muscle function. Although the examiner instructed the patient to contract the correct muscles, the measured strength of the muscle contraction by MOS digital examination could also be limited by the patient’s knowledge of how to isolate these pelvic muscles MOS is widely used in clinical practice, as it is easy to perform, inexpensive, and requires no special equipment. Furthermore, vaginal palpation is an effective aid in providing feedback to patients when they perform a pelvic floor muscle contraction [16]. The feasibility of this technique, however, is a strong point and allows for comparison with an examination technique that is a standard component of routine evaluation.

Ultrasound evaluation of the morphology of levator ani muscle is becoming more popular in urogynecology practice. The predictive value of morphology for functional assessment is a holds critical value in advancing knowledge in this field.