Inadequate Rectal Pressure and Insufficient Relaxation and Abdominopelvic Coordination in Defecatory Disorders

Abstract

Background and Aims:

Diagnostic tests for defecatory disorders (DDs) asynchronously measure anorectal pressures and evacuation and show limited agreement; thus, abdominopelvic-rectoanal coordination in normal defecation and DDs is poorly characterized. We aimed to investigate anorectal pressures, anorectal and abdominal motion, and evacuation simultaneously in healthy and constipated women.

Methods:

Abdominal wall and anorectal motion, anorectal pressures, and rectal evacuation were measured simultaneously with supine magnetic resonance defecography and anorectal manometry. Evacuators were defined as those who attained at least 25% rectal evacuation. Supervised (logistic regression and random forest algorithm) and unsupervised (k-means cluster) analyses identified abdominal and anorectal variables that predicted evacuation.

RESULTS.

We evaluated 28 healthy and 26 constipated women (evacuators comprised 19 healthy subjects and 8 patients). Defecation was initiated by abdominal wall expansion that was coordinated with anorectal descent, increased rectal and anal pressure, and then anal relaxation and rectal evacuation. Compared with evacuators, nonevacuators had lower anal diameters during simulated defecation, rectal pressure, anorectal junction descent, and abdominopelvic-rectoanal coordination (P<.05). Unsupervised cluster analysis identified 3 clusters that were associated with evacuator status (P<.01), i.e., 10 (83%), 16 (73%) and 1 (5%) evacuators in clusters 1, 2, and 3. Each cluster had distinct characteristics (eg, maximum abdominosacral distance, rectal pressure, anorectal junction descent, anal diameter) and correlates that were more (clusters 1–2) or less (cluster 3) conducive to evacuation. Cluster 2 had 16 evacuators (73%) and intermediate characteristics (eg, lower anal resting pressure and relaxation during evacuation [P<.05]).

Conclusions:

Women with DDs and a modest proportion of healthy women had specific patterns of anorectal dysfunction, including inadequate rectal pressurization, anal relaxation, and abdominopelvic-rectoanal coordination. These observations may guide individualized therapy for DD in future.

Graphical Abstract

LAY SUMMARY

Normal defecation entails a coordinated sequence of events: abdominal expansion, anorectal descent, and increased rectal and anal pressure followed by anal relaxation. Women with defecatory disorders have coordination or impaired rectal pressurization or anal relaxation (or both).

Background

Defecation requires coordination of propulsive forces (rectal pressure) and relaxation of the internal and external anal sphincters and the puborectalis muscle¹. Up to 50% of patients with chronic constipation have defecation disorders (DDs), which are characterized by impaired rectal evacuation. DDs can be attributable to inadequate rectal propulsive force, impaired anal relaxation, and structural disturbances (eg, rectoceles)². Unfortunately for patients and clinicians, diagnostic tests for DD may show limited agreement^3–5. For example, in one study of 125 patients with chronic constipation, 51% had dyssynergia by defecography, and of these 51%, only about half showed abnormal results with a balloon expulsion test or pelvic floor relaxation by surface electromyography³.

Indeed, the understanding of mechanisms underlying normal and impaired rectal evacuation and the tests used to diagnose DDs are subject to 4 critical limitations. First, no single criterion reference standard exists that can be used as a diagnostic test. Second, high-resolution manometry has confirmed a counterintuitive finding, namely that rectal pressure can be lower than anal pressure during evacuation in healthy people (ie, the rectoanal gradient is negative)^6–10. Perhaps this finding is at least partly explained by methodologic limitations because manometry is generally performed with the person in the left lateral position and with an empty rectum¹⁰. Third, anorectal pressures and rectal evacuation are measured separately with manometry and defecography (or a rectal balloon expulsion test), respectively. Hence, conflicting observations are frequent. Fourth, excessive abdominal strain is implicated in the pathogenesis of DD^{1, 11}. Pelvic floor biofeedback therapy has emphasized how patients with DDs need to improve abdominopelvic coordination during defecation^12–14, but no published studies to date have investigated abdominal wall motion or abdominopelvic-rectoanal coordination in normal defecation or in DDs.

Because of the limitations described above, this study aimed to assess the contribution of abdominopelvic and anorectal motion, anorectal pressures, and the coordination among these parameters during rectal evacuation in healthy and constipated women.

Materials and Methods

Study Design and Participants

This prospective study was approved by the Mayo Clinic Institutional Review Board. All subjects provided written, informed consent to participate in this study. We enrolled 30 women with functional constipation or constipation-predominant irritable bowel syndrome by Rome III criteria^{15, 16}. Participants did not have major systemic diseases, anorectal inflammation, or contraindications for magnetic resonance imaging (MRI), and they were not receiving medications that likely altered gastrointestinal motility. We also enrolled 30 healthy women without constipation as controls. They were matched by age and body mass index (BMI) to the women with constipation; they also could not have, anxiety, depression, or risk factors for pelvic floor trauma¹⁷.

Balloon Expulsion Test

Defecation was assessed by using a balloon expulsion test, as previously described^18–20. We measured the time required to expel a party balloon filled with 50 mL of water while the participant was in a seated position. An expulsion time greater than 60 seconds was considered prolonged.

Simultaneously Acquired Supine MR Defecography and Anorectal Manometry

Participant evaluations were conducted in a MR suite with a 1.5 T MRI system (GE Signa HDxt, software version HD16; General Electric). Participants self-administered 1 or 2 sodium phosphate enemas (C.B. Fleet Company, Inc), after which ultrasound gel (180 mL) was instilled into the rectum. An MRI-compatible anorectal manometry probe with 4-channel pneumohydraulic sensors was placed in the anorectum. The upper 2 sensors were located in the rectum, and the lower 2 sensors were situated 3 cm distally in the anal canal. The probe was connected to a software program (Mayo Clinic). Displacement of the probe during defecation was prevented by a clip attached to the catheter and the inner thigh.

Anorectal pressures were measured during MR image acquisition, while the participant was in the supine position. A sagittal oblique plane that bisected the anorectum was defined with planning images. Participants were coached to defecate²¹. During defecation, pressures and 2dimensional FIESTA (fast imaging employing steady-state acquisition) MR images were acquired every second for 90 seconds²². Typically, participants defecated twice for approximately 30 seconds per occasion, with a 30-second pause between maneuvers. We acquired images in real time with a slice thickness of 5 mm. Images were reconstructed in an oblique sagittal plane with a field of view of 36 cm, minimum time to echo (TE), flip angle of 60°, full-phase field of view, and matrix size of 192×160 (number of excitations [NEX]=1). We used a validated program (Pelvic Measure Program, Mayo Clinic 2017) to measure the rectal area, abdominal wall motion, anal length, anal diameter, and anorectal angle and descent; measurements were obtained at rest and during evacuation^{23, 24}. Abdominal wall motion (also termed abdominosacral distance) was defined as changes in the distance between the sacral promontory and abdominal wall musculature. In the analysis, we considered the maximum values obtained for abdominosacral distance, anal diameter, anorectal angle, and location of the anorectal junction; we additionally considered the minimum values obtained for rectal area and anal length. Rectal evacuation was summarized as the percent change in rectal area. We considered normal values for rectal emptying in asymptomatic women²¹ and defined “evacuators” as participants with 25% or more rectal emptying and “nonevacuators” as participants with less than 25% rectal emptying.

Comparison of Abdominopelvic-Rectoanal Coordination in Evacuators and Nonevacuators

We were interested in assessing coordination of the various parameters (described above) during the preparatory phase prior to defecation and early defecation. Thus, we measured temporal coordination between abdominal wall motion, anorectal motion, and anorectal pressures by evaluating cross-correlation of 2 variables (eg, the relative change in abdominosacral distance and rectal pressure) over 30s, i.e., in three consecutive temporal windows (each lasting 10s) after patients were instructed to defecate. Separate autoregressive, integrative, moving-average models were fitted to the 10-second time series of each pair of variables. The cross-correlation coefficients between the paired time series indicates the dependency between variables (eg, dependency between abdominosacral expansion and rectal pressure). (Intrinsic variability, which reflects the variability within a given time series, was not analyzed further.) The optimal time lag between the 2 variables was determined by measuring their correlations at 21 staggered 1-second intervals (from −10 to +10 seconds). At each interval, the strength of the correlation ranged from −1 to +1. The maximum correlation coefficient and the corresponding lag during this 30-second period was used for further analysis. For synchronous events, the correlation coefficient was highest at a lag of 0s. For asynchronous events, the correlation coefficient was greatest at a non-zero lag value. In a windowed time-series analysis with 10 samples, a correlation greater than 0.62 (ie, 1.96/√n, where n = number of samples) was considered statistically significant with an α value of 5%.

Statistical Analysis

We used the Wilcoxon rank sum test to compare abdominal wall motion, anorectal motion, and anorectal pressures. We also used it to assess the coordination between variables in evacuators versus nonevacuators, healthy versus constipated participants, and participants with normal and prolonged balloon expulsion time (BET). These comparisons used the summary parameters (eg, maximum rectal pressure) during the evacuation maneuver for which FIESTA MR images showed the most gel being expelled.

Thereafter, we used supervised (ie, logistic regression and random forest [RF] algorithm) and unsupervised bivariate and multivariate analyses (ie, k-means cluster analysis)²⁵ to identify factors that predicted the evacuator status of the study participant. In the univariate and bivariate logistic regression models with a generalized logit link function, the predictor variables were demographic characteristics (ie, age and BMI), change in anorectal pressures and motion during defecation, and the cross-correlations between variables. Because logistic models are ideally suited for analyzing linear relationships among variables, we also used the RF algorithm²⁶, accompanied by bootstrapping, to assess relationships among variables that may be nonlinear. The Boruta algorithm was used to identify the most useful predictor variables for the RF analysis²⁷. The importance value of a variable was calculated based on the loss of accuracy by random permutation of the variables. To assess whether the importance was significant, the Boruta algorithm compared the observed importance with that produced by spiked-in “shadow” variables, which were random permutations of real variables. Hence, the Boruta algorithm generally provides more power to identify features that predict a phenotype.

The unsupervised k-means cluster analysis incorporated abdominal wall motion, anorectal pressures, and coordination parameters but not rectal evacuation or BET. Posthoc nonparametric analyses were used to explore differences between the clusters.

The time series and k-means cluster analyses were conducted by using the tSeries packages in R 3.6.3 (R Core Team, 2014), and figures were produced with the package ggplot2²⁸. Other statistical analyses were performed with JMP (version 9.4, SAS Institute Inc).

Results

Clinical Features, Rectal Evacuation, and Balloon Expulsion Test

We enrolled 60 women in the study: 30 had constipation (“patients”) and 30 were control participants with normal (“healthy”) bowel function. In the patient group, the mean (SD) age was 36 (14) years and mean (SD), BMI was 27 (6). In the control group, the mean (SD) age was 38 (13) years and mean (SD) BMI was 26 (6) (Table 1).

Table 1.

Demographic Characteristics and Anorectal Variables

Characteristic or variable a	Balloon expulsion time			Rectal evacuation status
	Normal (n=41)	Prolonged (n=13)	P value	Evacuator (n=27)	Nonevacuator (n=27)	P value
Demographic characteristic
Age, y	38 (14)	37 (10)	0.35	38 (12)	37 (15)	0.90
Body mass index	27 (6)	23 (4) *	0.19	27 (5)	25 (5)	0.28
Pressure or dimension b
Maximum rectal pressure, mm Hg	73 (24)	58 (29)	0.16	84 (21)	54 (22) ***	<0.0001
Anal resting pressure, mm Hg	56 (27)	52 (14)	0.83	56 (29)	54 (19)	1.00
Anal relaxation (%)	7 (25)	−0.8 (21)	0.3	6 (29)	4 (18)	0.9
Rectoanal gradient, mm Hg	17 (29)	−7 (29)	0.046	28 (30)	−9 (27)	0.005
Maximum anorectal junction descent, mm	50 (20)	33 (22) *	0.01	57 (18)	35 (19) ***	<0.0001
Maximum abdominosacral distance, mm	85 (23)	66 (14) *	0.02	89 (21)	71 (21) ***	0.0007
Maximum anal diameter, mm	10 (6)	3 (4) ***	0.0002	13 (4)	3 (4) ***	<0.0001
Minimum anal length, mm	14 (7)	20 (11)	0.09	12 (5)	20 (9) ***	0.0007

^aData are presented as mean (SD).

^bDuring evacuation, unless stated otherwise.

^*:P≤0.05,

^**P ≤ 0.01, and

^***P ≤ 0.001

Among patients, 9 (30%) satisfied symptom criteria for irritable bowel syndrome with constipation and 21 (70%) met criteria for functional constipation. Bowel symptoms were less than 3 bowel movements per week (60%), incomplete evacuation (77%), straining (97%), hard stools (83%), anal blockage (70%), and anal digitation (23%). Healthy participants did not have bowel symptoms.

Technically satisfactory measurements of anorectal pressures and rectal emptying were obtained for 28 healthy women and 26 patients. Significantly more healthy women than patients (19/28 [73%] vs 8/26 [31%]; P<.001) were characterized as evacuators. Twenty-four of 28 healthy women (86%) and 17 of 26 patients (65%) had a normal BET. The rectal BET was shorter in evacuators than nonevacuators (mean [SD], 17 [35] vs 78 [81] seconds; P<.001). The rectal BET was negatively correlated with rectal emptying during evacuation (r=−0.54; P<.001).

Comparison of Pressures and Motion Between Evacuators and Nonevacuators

In evacuators, defecation was initially associated with expansion of the abdominal wall (ie, increased abdominosacral distance), anorectal junction descent, and increased rectal and anal pressure (Figure 1). Thereafter, the anal canal opened and became shorter, typically after rectal pressure approximated and then exceeded anal pressure. Rectal evacuation eventually occurred.

Figure 1.

Comparison of Rectoanal Variables and Magnetic Resonance Images in Evacuators and Nonevacuators. A-C, Average rectoanal and abdominal dimensions and pressures and rectal evacuation in evacuators (n=27) and nonevacuators (n=27). The black arrows point to the approximate onset of the 2 defecation maneuvers. A, Greater anal diameter and smaller rectal area (ie, more evacuation) were observed in evacuators. B, Anal resting pressures were similar for evacuators and nonevacuators. Among evacuators, rectal pressure markedly increased and then exceeded anal pressure, whereas among nonevacuators, the rectal pressure increment was much smaller. C, Abdominal expansion and anorectal junction descent were concurrent with the change in pressures shown in panel B. D-E, Representative images at baseline, onset of evacuation, maximum descent, and end of evacuation in an evacuator and a nonevacuator. The white dotted lines represent the abdominosacral distance, and the arrows point to the water-perfused manometry catheter. Abdominal expansion, anorectal descent, and rectal evacuation are greater in evacuators than in nonevacuators.

During evacuation, the maximum abdominosacral distance, maximum rectal pressure, maximum anorectal junction descent, and maximum anal diameter were greater in evacuators than in nonevacuators, and the minimal anal length was shorter (P<.001) (Table 1, Supplementary Figure 1). Among healthy women, the pattern of differences between evacuators and nonevacuators was similar to that of the overall patient cohort (Supplementary Table 1). With the exceptions of rectal pressure and anal length, the remaining variables were significantly different between participants who had a normal versus prolonged BET (P<.05) (Table 1).

Abdominal Wall-Rectoanal Coordination

The correlations between abdominal wall expansion versus rectal pressure and versus anorectal junction descent were significantly correlated (ie, greater than 0.62) in evacuators and nonevacuators (Supplementary Table 2). By comparison, the correlation coefficients for abdominal wall expansion and anorectal junction descent versus rectoanal gradient were not statistically significant and lower in evacuators and in nonevacuators, but nonetheless greater in evacuators than in nonevacuators.

The mean lag period between abdominal wall expansion and rectal pressure increase was 1.6 (95% CI, −0.1 to 3.3) seconds in evacuators and 1.2 (95% CI, −0.5 to 2.8) seconds in nonevacuators (Supplementary Table 3). The 95% confidence interval included 0 for both groups, which suggests that abdominal wall expansion and rectal pressurization occurred simultaneously. The anorectal junction descent and increase in rectal pressure similarly were somewhat simultaneous. In contrast, the mean lag period for the correlation between abdominal wall expansion and anorectal junction descent was 0.2 (95% CI, −1.6 to 2.0) seconds in evacuators and 2.3 (95% CI, 0.7–3.9) seconds in nonevacuators. For this correlation, the 95% confidence interval did not include 0 only for nonevacuators, which suggests that anorectal junction descent occurred after abdominal expansion in nonevacuators but was nearly synchronous in evacuators.

Logistic Regression

In the univariate models, maximum anal diameter, maximum rectal pressure, maximum anorectal junction descent, minimum anal length, and maximum abdominosacral distance were significantly associated with evacuator status, as were correlations between select variables (Figure 2A, Table 2). In the univariate analysis, several variables were associated with a significantly increased probability of being an evacuator: maximum anal diameter (odds ratio [OR], 1.61; 95% CI, 1.27–2.06), maximum rectal pressure (OR, 1.07; 95% CI, 1.03–1.11), maximum anorectal junction descent (OR, 1.06; 95% CI, 1.03–1.11), minimum anal length (OR, 0.86; 95% CI, 0.78–0.94), and maximum abdominosacral distance (OR, 1.04; 95% CI, 1.01–1.08). A mean OR of 1.61 indicated that a 1-mm increase in maximum anal diameter during evacuation was associated with a 61% increase to being an evacuator. Among the univariate models, the area under the receiver operating curve (AUROC) was greatest (0.94) for maximum anal diameter. In the bivariate models (Figure 2B), maximum anal diameter was combined with different variables and AUROC values were calculated: combined with minimum anal length, AUROC=0.97; with maximum rectal pressure, AUROC=0.96; and with maximum abdominosacral distance, AUROC=0.95. Similarly, the maximum rectal pressure and selected coordination correlation coefficients also predicted evacuator status in the univariate and bivariate models (Figure 2B).

Figure 2.

Logistic regression models for discriminating between evacuators and nonevacuators. A, Area under the receiver operating curve (AUROC) values for univariate and bivariate models. Among the univariate models (labeled squares), the maximum anal diameter (AUROC=0.94) and maximum rectal pressure (AUROC=0.84) were best at discriminating between evacuators and nonevacuators. B, Comparison of AUROC among models. The bivariate models had greater predictive value than the univariate models. For example, combining maximum anal diameter and minimum anal length yielded an AUROC of 0.97; combining maximum anal diameter with maximum rectal pressure yielded an AUROC of 0.96.

Table 2.

Univariate and Bivariate Analysis of Risk Factors for Evacuators vs Nonevacuators

Variable	Univariate models		Bivariate models
			Model 1	Model 2	Model 3	Model 4	Model 5
	OR (95% CI)	AUC	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)
Maximum anal diameter, mm	1.62*** (1.27–2.06)	0.94	1.63* (1.23–2.16)	1.68** (1.19–2.39)	1.64* (1.26–2.13)
Maximum rectal pressure, mm Hg	1.07*** (1.03–1.11)	0.84		1.07 (0.99–1.16)		1.09*** (1.04–1.14)	1.07** (1.03–1.11)
Maximum anorectal junction descent (mm)	1.07*** (1.03–1.11)	0.81
Maximum abdominosacral distance, mm	1.04** (1.01–1.08)	0.76			0.99 (0.95–1.03)
Minimum anal length, mm	0.86** (0.79–0.95)	0.76	0.86 (0.74–1.02)
Correlation between abdominosacral distance and rectoanal gradient	1.72 (0.86–3.44)	0.72				2.63** (1.16–5.95)
Correlation between anorectal junction descent and rectoanal gradient	3.05** (1.07–8.66)	0.73					2.70 (0.89–4.81)

Abbreviations: AUROC, area under the receiver operating curve; NA, not applicable; OR, odds ratio.

^*:P≤0.05,

^**P ≤ 0.01, and

^***P ≤ 0.001

RF Analysis

Using the RF classifier with features selected by the Boruta algorithm (Supplementary Table 4), we determined that the most important variables that discriminated between evacuators and nonevacuators were maximum anal diameter, rectal pressure, anorectal junction descent, abdominosacral distance, and minimum anal length and the correlation between abdominosacral distance versus rectoanal gradient.

We used the RF analysis as a sensitivity analysis to assess interactions and identify combinations of variables that affected rectal evacuation. If the maximum anal diameter was greater than 12 mm and the maximum rectal pressure was greater than 50 mm Hg, evacuation always occurred. However, if the maximum anal diameter was less than 12 mm, then evacuation was unlikely unless rectal pressure exceeded 75 mm Hg. Patients with an open anal canal but a suboptimal rectal pressure may or may not evacuate. Similar findings were observed in the other models. Therefore, an anal diameter less than 12 mm, rectal pressure less than 50 mm Hg, anorectal junction descent less than 25 mm, and anal length greater than 15 mm were associated with a lower probability of evacuation (Figure 3).

Figure 3.

Random Forest Analysis to Predict the Probability of Being a Nonevacuator. In all 3 panels, maximum anal diameter is represented on the x-axis because it was the strongest predictor of evacuation in the univariate logistic regression model and the Boruta parameter selection process. The blue colors indicate a lower probability of being a nonevacuator (ie, more evacuation). A maximum anal diameter >12 mm (upper, mid, and lower panels), rectal pressure >50 mm Hg (upper panel), anorectal junction descent >25 mm (mid panel), and minimum anal length <15 mm (lower panel) were associated with higher probability of evacuation.

k-Means Cluster Analysis

The k-means cluster analysis uncovered clusters defined by motion, pressure, and coordination variables (Figure 4A). The BET and rectal evacuation were not included in this analysis. An elbow plot suggested that 3 clusters optimally separated the participants (Figure 4B). However, the clusters that were identified were associated with evacuator status (P<.001) and duration of BET (P<.01) (Figure 4C). Clusters 1 and 2 comprised mostly evacuators (n=10 (83%) and n=16 (73%), respectively), and cluster 3 comprised mostly nonevacuators (with only one evacuator [5%]).

Figure 4.

Cluster Analysis. A, The panel depicts the 3 clusters, the evacuators (E) and nonevacuators (NE) in each cluster, and the centroid of each cluster. B. This elbow plot shows that the optimal number of clusters, which is three. C, Composition of participants in each cluster. In clusters 1 and 2, a majority were evacuators. In cluster 3, all but one were nonevacuators. These clusters were also associated with balloon expulsion time (BET).

For each cluster, we identified the centroid, and then for each centroid, we determined the numeric values of the anorectal variables (Figure 5, upper panel). In clusters 1 and 2, the anorectal variables (eg, maximum abdominosacral distance, rectal pressure, and anal diameter) and correlations showed values that generally seemed consistent with rectal evacuation. Clusters 1 and 2 were generally similar. However, anal resting pressure and relaxation during evacuation were lower in cluster 2 than cluster 1. By contrast, cluster 3 was characterized by prototypical features of impaired rectal evacuation. All 7 individual anorectal variables, the correlation between abdominosacral expansion and rectoanal gradient, and the correlation between anorectal junction descent and the rectoanal gradient were different between clusters 1 and 3. A diagram showing the relationship of clusters in terms of anorectal pressures and dimensions and the correlations between anorectal variables confirmed intermediate values for cluster 2 (Figure 5, lower panel).

Figure 5.

Upper panel. Values for rectoanal variables, corresponding to the centroids of clusters 1–3. Cluster 1 was characterized by normal rectal propulsion, anal relaxation, abdominopelvic motion, and coordination. Compared to cluster 1, cluster 2 was characterized by reduced rectal propulsion with anal contraction, less abdominopelvic motion, and normal coordination. The key features of cluster 3 were least rectal propulsion, reduced anal relaxation, least abdominopelvic motion, and reduced coordination. In clusters 1 and to a lesser extent cluster 2, the anorectal features (eg, maximum abdominosacral distance, rectal pressure, anal diameter) and correlations were more conducive to rectal evacuation than in cluster 3. Lower panel. Significant differences in rectoanal variables (left panel) and correlation coefficients between clusters (right panel) are shown. These differences were more pronounced in pressures and dimensions (left panel) than the correlations. For example, the anal pressure at rest and anal relaxation during evacuation but not correlation coefficients were significantly different between clusters 1 and 2. Clusters 1 and 2 were very different from cluster 3. By comparison, differences between cluster 1 and 2 were modest. The symbol “>” indicates greater; “<,” less or smaller. E indicates evacuators, NE, nonevacuators.

Discussion

Anorectal dysfunction is implicated in DDs; for example, inadequate rectal propulsive forces or impaired anal relaxation can prevent normal defecation^{1, 29, 30}. Measurements of pressures or motion during normal defecation suggest that rectal pressure must be greater than anal pressure, but in some healthy people, anal pressure is greater than rectal pressure during evacuation. This variability limits the use of high-resolution anorectal manometry to diagnose DD and undermines our understanding of the mechanisms of normal defecation and DD^{7, 9}. No prior studies have investigated abdominal wall motion or its coordination with pelvic floor motion during defecation. By simultaneously investigating abdominal wall expansion, anorectal motion, and anorectal pressures, the coordination among these variables, and rectal emptying during evacuation, we have gained new insights and a more refined, quantitative, understanding into the role of abdominal wall expansion, changes in the anal canal, rectal propulsive forces, and abdominopelvic-rectoanal coordination, which is a broad term that refers to coordination among all these events during normal defecation, and we have characterized how they may disturbed in DDs.

For evacuators, abdominal wall expansion was one of the earliest events in defecation. As confirmed by the correlation analysis, this expansion was coordinated with anorectal junction descent and increased rectal and anal pressure. Rectal pressure then exceeded anal pressure, thereby establishing a positive rectoanal gradient that facilitated evacuation. In the nonparametric RF analysis and regression models, the maximum rectal pressure and anal diameter were the most useful variables for discriminating between evacuators and nonevacuators; anorectal junction descent and the minimum anal length were also useful. . The maximum abdominosacral distance and coordination correlation coefficients also were significant predictors in the regression model. Among evacuators, the average maximum anal diameter was 1.3 cm, which was similar to the measurement observed with barium defecography (1.6 cm)³¹. In the univariate analysis, maximum anal diameter and rectal pressure during evacuation were the 2 most discriminatory variables, which confirms that an adequate propulsive force is necessary for normal defecation²⁹. Indeed, the maximum rectal pressure was considerably greater in evacuators than nonevacuators (mean [SD], 84 [21] vs 54 [22] mm Hg). Abdominal expansion (ie, increased abdominosacral distance) was correlated with anorectal junction descent and with rectal pressure (Table 2), which strongly suggests that these events occur concurrently. Diaphragmatic descent, which is known to be accompanied by abdominal expansion and increased abdominal pressure^{32, 33}, may be responsible for these events and may at least partly explain increased rectal pressure during defecation.

Even among evacuators, anal pressure initially increased during defecation. This increase in pressure is inconsistent with the concept of anal relaxation as a feature of abnormal defecation. The pressure increase is volitional, precedes the opening of the anorectal junction (is not transmitted from the abdomen), and occurs during defecation. It differs from the anal sensorimotor response that is evoked by rectal distention and the desire to defecate³⁴; rather, it resembles the contraction of the external anal sphincter during coughing, straining, and the Valsalva maneuver, which are somatic responses that preserve continence¹¹. Although the higher control of defecation is poorly understood, this early anal contraction conceivably is a somatic accompaniment to straining during defecation. This contractile response subsequently is replaced by anal relaxation, which is a visceral response that perhaps is evoked by mucosal stimulation and/or distention of the distal rectum (ie, rectoanal inhibitory reflex) and/or marked urgency or is associated with activation of the pelvic organ stimulation center³⁵.

In contrast to the logistic regression models that we used, the unsupervised cluster analysis did not include rectal evacuation or BET status. Nonetheless, the 3 clusters appear to have face validity because they were able to discriminate between evacuators and nonevacuators and between participants with normal or prolonged BET. In contrast to earlier studies, which characterized patients by using measurements from left lateral^{29, 30} or upright manometry¹⁰ or defecography²², the current studies were performed with manodefecography with a filled rectum. In cluster 1, which consisted only of evacuators, 9 of the 11 women (82%) were healthy and all had a normal BET. This cluster was characterized by sufficient abdominal wall expansion, perineal descent, and anal opening, an adequate rise in rectal pressure, and high correlation coefficients among parameters, which suggests normal abdominopelvic-rectoanal coordination. In contrast to cluster 1, 10 of 22 (45%) women in cluster 2 had constipation. These women had a higher anal resting pressure and more relaxation during evacuation. Cluster 3 was characterized by multiple abnormalities (ie, impaired perineal descent, anal relaxation, rectal propulsive forces, and coordination); 19 of 20 were nonevacuators. The correlation between abdominosacral expansion and rectoanal gradient was significantly lower in cluster 3 than cluster 1. Taken together, these finding suggest that our cohort spanned the spectrum from normal (cluster 1) to intermediate (cluster 2) to abnormal (cluster 3) anorectal function.

In clusters 1 and 2, most patients were evacuators; the few nonevacuators were situated at the periphery of these clusters. This suggests that even within a cluster, the % evacuation is determined by the interaction of several variables. The RF analysis quantified the probability of being a nonevacuator (Figure 3) and showed that it was more likely (probability >75%) when the anal diameter was 0 to 4 mm, possible (probability of 25%−75%) when the diameter was 4 to 12 mm, and unlikely (probability <25%) when the diameter was wider than 12 mm. Further, across the range of values for anal diameter, the magnitude of the y-axis variable influenced the probability of evacuation. For example, when the anal diameter was between 4 and 12 mm, the probability of inadequate evacuation was more than 50% for rectal pressures up to 50 mm Hg and 50% or lower for rectal pressures of 75 mm Hg or greater. These data suggest that greater rectal pressure can offset inadequate widening of the anal canal.

These findings may have future clinical implications. At present, pelvic floor biofeedback therapy for DDs is often not individualized to the specific underlying disturbance, but limited evidence now indicates that anorectal pressure topography may predict the response to biofeedback therapy in DDs³⁶. Thus, anorectal pressure topography conceivably could be used to facilitate individualized biofeedback therapy for DD. For example, among nonevacuators in cluster 2, therapy could focus on improving the coordination between abdominal wall expansion (measured with abdominal wall plethysmography) and anal relaxation (measured with anal electromyography or manometry); in cluster 3, therapy could address global abdominopelvic-rectoanal disturbances. The imaging sequences for MR defecography are widely available on clinical MRI scanners; technical standards have been published³⁷. Nevertheless, some centers do not offer this test, perhaps partly because insurance reimbursement is variable, despite evidencebased recommendations issued by the American Gastroenterological Association, American College of Gastroenterology, and American College of Radiology that endorsed this test in patients when clinically appropriate^{2, 38, 39}. Like other anorectal tests, our assessments also reported false-positive findings in some healthy controls. Hence, test results need to be interpreted in conjunction with clinical findings. At present, MR defecography reports in clinical practice and research are limited to anorectal motion. Adding the abdominosacral distance to these reports should heighten awareness of impaired abdominal wall expansion in DD. MR manodefecography can be performed with MR-compatible water-perfused pressure sensors but not with solid-state, high-resolution manometry sensors (Manoscan, Medtronic). By contrast, barium-defecating manodefecography could be performed with solid-state pressure sensors and it is less expensive and more widely available than MR defecography. However, barium defecography has lower spatial resolution, especially when measuring abdominosacral distance, and in contrast with MR manodefecography, it also entails radiation exposure.

We acknowledge the strengths and limitations of our study design. This study was conducted by using state-of-the-art techniques in a cohort of healthy controls and patients who were matched by age and BMI. MR manodefecography eliminated several limitations associated with isolated high-resolution anorectal manometry. We used water-perfused pressure sensors, which were less expensive (although probably less convenient) than the fiberoptic pressure sensors that are MRI compatible. However, the experiments were conducted only in women. There are several differences between normal defecation and MR defecography. Although supine MRI and seated defecography have comparable accuracy for diagnosing posterior floor disorders⁴⁰, some women may find it challenging to evacuate in the supine position²¹ or to defecate in a test environment (or both). The inadequate rectal evacuation (BET >60 seconds) observed in 15% of healthy women may indicate the presence of asymptomatic pelvic floor dysfunction⁹. Although the rectal area measured with this program has been validated with rectal volume measurements²⁴, it is conceivable that area measurements may have underestimated emptying in some participants.

In summary, normal defecation is characterized by abdominal expansion, increased rectal and anal pressure, and then anal relaxation. Women with DDs and a modest proportion of healthy women have specific patterns of anorectal dysfunction that are characterized by inadequate rectal pressurization, anal relaxation, or abdominopelvic-rectoanal coordination (or a combination thereof). A cluster analysis of various rectoanal and abdominopelvic motion and pressure variables illustrates the phenotypic heterogeneity in DD.

WHAT YOU NEED TO KNOW?

Background and context:

Rectoanal pressures and evacuation typically are measured asynchronously limiting our understanding about the mechanisms of normal defecation and the pathogenesis and diagnosis of defecatory disorders (DDs).

New findings:

Normal defecation was characterized by abdominal expansion coordinated with rectoanal descent and increased rectal and anal pressure followed by anal relaxation. Women with DDs had specific patterns of rectoanal dysfunction, characterized by impaired abdominopelvic-rectoanal coordination or impaired rectal pressurization or anal relaxation (or both).

Limitations:

We evaluated only women in this study. Magnetic resonance manodefecography evaluations were conducted with participants in the supine position.

Impact:

These findings provide new insights into the mechanisms of normal defecation (eg, primacy of abdominal expansion and abdominopelvic-rectoanal coordination, and increased rectal and anal pressure followed by anal relaxation) and into the pathogenesis of DDs.

Abbreviations

AUROC

area under the receiver operating curve

BET

balloon expulsion time

BMI

body mass index

DD

defecation disorder

FIESTA

fast imaging employing steady-state acquisition

MR

magnetic resonance

MRI

magnetic resonance imaging

NEX

number of excitations

OR

odds ratio

RF

random forest

TE

time to echo