Significance of Pelvic Floor Muscles in Anal Incontinence

Transcription

1 GASTROENTEROLOGY 2002;123: Significance of Pelvic Floor Muscles in Anal Incontinence XOSE FERNÁNDEZ FRAGA, FERNANDO AZPIROZ, and JUAN R. MALAGELADA Digestive SystemResearch Unit, University Hospital Vall d Hebron, Autonomous University of Barcelona, Barcelona, Spain Background & Aims: The pathophysiology of anal incontinence may be elusive using current parameters. Our aim was to establish the role of the levator ani in anal continence. Methods: In 53 patients with anal incontinence, 30 with constipation as disease controls, and 15 healthy controls, we evaluated incontinence severity by a 0 12 scale, anorectal function by standard manometric tests, and levator ani contraction by a perineal dynamometer. Results: Patients with incontinence exhibited various physiologic abnormalities ( per patient), but multiple regression analysis showed that levator ani contraction was the independent variable with strongest relation to the severity of incontinence (R 0.84; P < ), as well as a predictive factor of the response to treatment (R 0.53; P < 0.01). Furthermore, in contrast to other physiologic parameters, clinical improvement in response to treatment ( score vs score pre; P < 0.001) was associated with a marked and significant strengthening of levator ani contraction ( g vs g pre; P < 0.05). Conclusions: We have shown the importance of levator ani failure in understanding the etiology of anal incontinence and in predicting response to treatment. The function of the anorectum is to preserve anal continence, but the specific mechanisms involved are not well defined. Anal incontinence is an important problem both to the patient, for its dramatic impact on the quality of life, and for the society, for its tremendous socioeconomic burden. 1 3 However, treatment of incontinence is largely empiric. Muscular rehabilitation has been proven an effective form of therapy, 4 8 yet operatingvia mechanisms that are not well understood. For instance, the clinical improvement produced by biofeedback treatment does not appear to be accompanied by objective improvement of the external anal sphincter function. This is surprisingbecause the striated muscular component of the anus is, theoretically, a main target of this form of therapy. 7 9 Several factors have been implicated in anal continence: the reservoir function of the rectum, the muscular closure of the anal canal, and the neural control mechanisms, both intrinsic and extrinsic The muscular apparatus involves the internal sphincter, which is a smooth muscle cylinder surroundingthe anal canal and controlled by the intrinsic myenteric innervation, and the striated musculature, which is modulated by the extrinsic innervation and under voluntary control The striated component includes the external anal sphincter, which is a second muscular cylinder surroundingthe internal anal sphincter, and the levator ani. 11,15 17 Of the latter, the most prominent component is the puborectalis, with 2 well-developed parallel fascicles that insert anteriorly in the pubis and fusion together behind the rectum at the top of the anal canal The traction exerted by the puborectalis slingmaintains the anorectal angulation. 11,17 The other components of the levator ani are 2 thin sheets of muscle that complete the pelvic diaphragm, namely the iliococcygeus and the pubococcygeus, although different anatomic arrays have been described Indirect evidence suggests that the levator ani is a major contributor to anal continence. For instance, complete sphincteric section duringsurgical treatment of anal fistulas may not compromise continence, as longas the levator ani muscle is preserved. 16 Our aim was to produce direct evidence of the role of the levator ani muscles on anal continence. In a relatively large group of patients with anal incontinence we measured the severity of incontinence both before and after biofeedback treatment by a scale, anorectal function by a battery of tests, and levator ani contraction by a perineal dynamometer. We correlated first, the severity of incontinence with the various physiologic parameters measured, and second, the clinical response to treatment with the specific physiologic changes. Materials and Methods Participants Fifty-three patients with anal incontinence (45 women, 8 men; age range, years), 30 patients with constipation as disease controls (23 women, 7 men; age range, Abbreviation used in this paper: NS, not significant by the American Gastroenterological Association /02/$35.00 doi: /gast

2 1442 FERNÁNDEZ FRAGA ET AL. GASTROENTEROLOGY Vol. 123, No. 5 Table 1. Incontinence Score Frequency Type of incontinence Gas Liquid Solid Never /mo Monthly a Weekly b Daily c NOTE. Global incontinence score (0 12) was calculated by summation of gas, liquid, and solid scores. a Monthly: 1/mo to 1/wk. b Weekly: 1/wk to 1/d. c Daily: 1/d years), and 15 healthy subjects (12 women, 3 men; age range, years) were included in the study. Both incontinence and constipation were defined by usingpreviously established criteria. 18,19 The study protocol was approved by the Institutional Review Board of the Vall d Hebron University Hospital. All participants gave written informed consent before the study. Clinical Evaluation of Anal Incontinence The clinical severity of incontinence was measured by an incontinence scale, as follows (Table 1): incontinence to gas, liquids, and solids was respectively scored from 0 to 4 on 3 separate frequency scales as 0 (never), 1 (less than once per month), 2 (once or more per month, but less than once per week), 3 (once or more per week, but less than once per day), and 4 (once or more per day). A cumulative incontinence score (ranging from 0 to 12) was calculated by summation of the individual gas, liquid, and solid scores. The duration of incontinence from the appearance of the first symptoms to the time of the study was also recorded. Measurement of Levator Ani Contraction The voluntary muscular contraction of the levator ani was measured by usinga perineal dynamometer (Figure 1). With the patients positioned in lateral decubitus, we measured the traction exerted by the levator ani on an intrarectal balloon catheter. The latex balloon was inflated with a fixed volume of water and was connected by a nondistensible rope to an electronic isometric dynamometer (series BC 300, DS Europe, Milan, Italy). The external dynamometer was positioned perpendicular to the anal canal axis by means of an articulated support, which was then fixed in relation to the bony structures of the pelvic floor. The support consisted of 3 articulated arms made of methacrylate. The dynamometer was enclosed in the central arm ( cm). The anterior ( cm) and posterior arms ( cm) were adjusted and fixed in each subject to be in apposition to the anteroinferior aspect of the pubis and the posteroinferior aspect of the sacrococcigeal junction, respectively. The arms were articulated by slidingslots and adjustable screws. The support was maintained in place against the pelvic bones by manual pressure. Askingthe patients to relax, the rope connectingthe balloon was then fixed by an adjustable screw to the isometric dynamometer at the desired basal tension level. Phasic contraction of the levator ani was measured by askingthe subjects to produce a maximal anal squeeze. Before each study the dynamometer was calibrated by applying100 and 500 gweights. Data were recorded on a computerized polygraph (Mark II Biological Monitoring, Sicie, Barcelona, Spain). Evaluation of Anorectal Function Specific aspects of anorectal function were evaluated by a battery of standard tests with the patients positioned in lateral decubitus, as follows. Anal manometry. Contraction of anal sphincters was evaluated by usinga low-compliance manometric perfusion system (0.1 ml/min perfusion rate) and a 4 radial-lumen polyvinyl catheter (2.4 mm OD, Arndorfer Medical Special- Figure 1. Perineal dynamometer. The contraction of the levator ani was measured as the traction exerted on an intrarectal balloon catheter, connected to an isometric dynamometer, fixed by an articulated support to the pelvic bones. Scanned picture of the articulated support shown.

3 November 2002 LEVATOR ANI AND ANAL CONTINENCE 1443 ties, Greendale, WI) by a stationary pull-through technique at 1-cm steps. The tonic contraction of the internal anal sphincter was evaluated by the basal anal pressure (pressure referenced to intrarectal pressure), and the phasic contraction of the external anal sphincter by the maximum squeeze pressure (pressure increment from basal). 12,13 Neural reflexes. Reflex responses were evaluated by usinga 5 lumen polyvinyl catheter (4.8 mm OD, Arndorfer Medical Specialties) with manometric ports 1 cm apart and with a latex balloon at the tip fixed at 5 mm from the distal port. The rectoanal inhibitory reflex was tested by inflation of the rectal balloon with air by the followingprocedure: phasic rectal distensions of 10-second duration were performed at 1-minute intervals and in 10-mL increments, while measuring the reflex relaxation of the internal anal sphincter (basal anal pressure drop). This is an intrinsic reflex driven by fibers in the myenteric plexus. 11,13,20,21 The cough reflex was triggered by askingthe patients to cough, and both the abdominal compression (intrarectal pressure peak) and the reflex contraction of the external anal sphincter (anal pressure increment) were measured. This is a sacral reflex driven by the pudendal nerves. 22,23 Abnormal cough reflex was defined as anal pressure peak lower than both the intra-abdominal pressure peak and the voluntary squeeze pressure. Rectal compliance and sensitivity. An oversized polyethylene bag(600-ml capacity, 28-cm maximal perimeter) connected to a double-lumen polyvinyl tube (12 F, Argyle, Sherwood Medical, St. Louis, MO) was introduced into the rectum. Rectal distension was produced by step-wise inflation of air in 20-mL increments every 15 seconds up to the level of discomfort while measuringintrarectal pressure and subjective sensations (first perception, urge, and discomfort). In each subject, rectal compliance 10 13,24 was evaluated by the intrarectal pressure at 80-mL distendingvolume, which was the highest inflation level tolerated by all patients and healthy subjects. Treatment of Anal Incontinence Anal incontinence was treated by means of a biofeedback technique. 4 8 Intrarectal and anal pressures were recorded and displayed on a monitor in view of the patients. 8 Under visual control, the patients were instructed to squeeze for 5 seconds, tryingto increase anal pressure without contracting the abdominal musculature (i.e., without intrarectal pressure increments). Each session lasted minutes. The patients were then instructed to exercise twice daily for 10 minutes, alternating5-second squeeze and 10-second rest intervals. Reinforcement sessions were scheduled at 3-month intervals. Specific Procedures and Design Preliminary studies. In 14 patients (10 incontinent, 4 constipated) we investigated the influence of intrarectal balloon volume and of the basal tension applied to the balloon (traction set on the balloon catheter) on levator ani measurements by the dynamometer. In each subject, the force of levator ani contraction was measured with 25 and 40 ml of Figure 2. Example of levator ani contraction. The peak force of 3 consecutive contractions was averaged for calculations. Figure corresponds to incontinent patient. water within the intrarectal balloon and each volume level was tested at 50, 100, and 250 gbasal tension, respectively. In each test, subjects were asked to perform 3 contractions of 5-second duration separated by 10-second restingintervals. Each of the 6 experimental conditions (25- and 40-mL intraballoon volume both at 50, 100, and 250 gbasal tension) were tested in the 14 subjects in random order, and at least a 1-minute rest period was allowed between tests. Main Studies These were performed in 42 nonselected incontinent patients, 24 constipated patients, and 15 healthy subjects. Basal evaluation. In 30 patients with incontinence (24 women, 6 men; mean age, 61 2 years), 24 with constipation (19 women, 5 men; mean age, 46 3 years), and 15 healthy subjects (12 women, 3 men; mean age, 56 3 years) we measured: (1) levator ani contraction by usingthe perineal dynamometer (with 25 ml of water within the intrarectal balloon and at a basal tension of 100 g) as the average peak force of 3 contractions of 5-second duration separated by 10-second intervals (Figure 2), (2) anorectal function by means of the battery of tests described earlier, and (3) the clinical severity of incontinence by usingthe incontinence scale. Effect of treatment. In 25 incontinent patients, the measurements (levator ani contraction, anorectal function, and incontinence score) were repeated 6 months after starting biofeedback treatment (immediately before the second reinforcement sessions of biofeedback), and in 23 of them additional measurements were performed 3 months after starting the treatment. No treatment group. To exclude putative effects of enrollment into the study on test performance or subsequent clinical evolution, in an additional group of 12 nonselected incontinent patients, who were on the waitinglist for biofeedback, we repeated the same tests after a 3-month interval without treatment.

4 1444 FERNÁNDEZ FRAGA ET AL. GASTROENTEROLOGY Vol. 123, No. 5 Ancillary Studies Reproducibility studies. In 15 patients (11 incontinent, 4 constipated) we performed a second measurement of levator ani contraction after a 5 10 day interval, followingthe same procedure as in the main studies. Rheologic properties of the levator ani. In 23 patients with incontinence (20 women, 3 men; mean age, 61 2 years), 17 patients with constipation (13 women, 4 men; 61 2 years), and 13 healthy subjects (11 women, 2 men; 55 3 years) the rheologic properties of the levator ani were compared with those of the external anal sphincter by using2 different paradigms: sustained squeeze for as longas possible, and intermittent, brief squeezes of 2 seconds duration at 2-second restingintervals (guided by a visual signal) during 2 minutes. With each paradigm we measured in separate tests the contraction of the levator ani by the perineal dynamometer and of the external anal sphincter by manometry. Data Analysis and Statistical Analysis In each group of subjects we calculated the mean values ( SE) of the different parameters measured. In the preliminary studies, measurements of levator ani contraction were analyzed by a 2-way analysis of variance (ANOVA) with contrasts to show the simple effects of balloon volume and basal tension. A between-within univariate ANOVA was used to compare means of levator ani contraction at different balloon volumes and basal tension levels. In the reproducibility studies, the Bland-Altman plot was used to assess the repeatability of levator ani contraction measurements. 25 The Kolmogorov-Smirnov test was used to test normality of data distribution. Comparisons of parametric, normally distributed data were performed by the Student t test, paired tests for intragroup comparisons, and unpaired tests for intergroup comparisons; otherwise the Wilcoxon signed-rank test was used for paired data, and the Mann-Whitney U test was used for unpaired data. Correlations of paired data were examined by linear regression analysis. Multiple linear regression analysis was used to predict the dependent variable from the independent variables. Because no differences in the physiologic parameters tested were found between healthy and constipated controls, both groups were pooled to calculate the 95 percentile values of the parameters measured, and then we calculated the proportion of incontinent patients below these normal ranges (i.e., patients with impaired values). By using these cut-off values, the physiologic parameters were dichotomized (normal 0, abnormal 1), and from the scoringof these variables, a single prediction score for incontinence (no 0, yes 1) was established for each patient, to evaluate each individual patient in the database. The sensitivity and specificity of the incontinence prediction score was estimated by means of a receiver operatingcharacteristic analysis. 26 Contingency tables were analyzed by the Fisher exact or the 2 test, as pertinent. Results Preliminary Studies Neither the volume within the rectal balloon nor the basal tension exerted on the balloon catheter influenced the measurements of levator ani contraction by the perineal dynamometer: the force of levator ani contraction measured with 25 ml within the intrarectal balloon was g, g, and gapplying 50, 100, and 250 gbasal tension, respectively; correspondingvalues of levator ani force measured with 40 ml within the rectal balloon were g, g, and g, respectively. The values tended to be higher with higher basal tension, but the differences were small and statistically nonsignificant. Based on these preliminary studies, all subsequent measurements were performed by using25 ml within the rectal balloon and 100 gbasal tension. Anorectal Physiology in Incontinence Patients with incontinence, as compared with controls, had markedly impaired anal basal pressures (44 3mmHgvs.68 3mm Hgin healthy controls and 76 4 mm Hgin constipated controls; P.001 for both), impaired squeeze pressures (47 7mmHgvs mm Hgin healthy controls and 130 9mm Hgin constipated controls; P for both), increased rectal perception (first perception at 40 4mL vs ml in healthy controls and 53 6mLin constipated controls; urge to defecate at 68 4mLvs ml in healthy controls and ml in constipated controls; discomfort at ml vs ml in healthy controls and ml in constipated controls; P 0.05 for all), and reduced rectal compliance (26 2 mm Hgintrarectal pressure at 80 ml intrarectal volume vs mm Hgin healthy controls and 18 2 mm Hgin constipation; P 0.05 for both). Furthermore, 43% of patients with incontinence exhibited an abnormal cough reflex (vs. 0% in healthy and constipated controls; P 0.05). All subjects had normal rectoanal inhibitory reflexes. By usingthe perineal dynamometer we also showed a significantly impaired contraction force of the levator ani muscle in incontinent patients ( gvs gin healthy controls and gin constipated controls; P 0.01 for both). Takingas a reference the 95th percentile values of the parameters in both healthy and constipated controls, 77% of the incontinent patients (but only 8% of controls; i.e., 92% specificity) had impaired basal anal pressure, 73% had impaired squeeze pressure (92% specificity), 44% had increased perception of intrarectal volumes

5 November 2002 LEVATOR ANI AND ANAL CONTINENCE 1445 Table 2. Correlation Matrix of Clinicophysiologic Parameters for 30 Incontinent Patients Clinical score Levator ani force Squeeze pressure Basal pressure Rectal discomfort Rectal compliance Clinical score - 70 a 54 a 7 a 2 a 4 a Levator ani force 0.84 b - 41 a 1 a 1 a a c Squeeze pressure 0.74 b 0.64 b - 6 a 1 a 2 a c c Basal pressure 0.26 b 0.11 b 0.26 b - 12 a 3 a 0.15 c 0.55 c 0.16 c Rectal discomfort 0.13 b 0.10 b 0.10 b 0.35 b - 23 a 0.52 c 0.58 c 0.61 c 0.08 c Rectal compliance d 0.21 b b 0.14 b 0.18 b 0.48 b 0.29 c 0.99 c 0.46 c 0.36 c 0.01 c - a Coefficient of determination (R 2 ) expressed as percentage. b Pearson correlation coefficient (R). c Significance (P). d Rectal pressure at 80 ml intrarectal volume. (93% specificity), 33% had reduced rectal compliance (93% specificity), and 60% had impaired levator ani contraction (92% specificity). Consideringthese 5 parameters plus the cough reflex, incontinent patients had a mean number of abnormalities of per patient. By usinga receiver operatingcharacteristic analysis, we found one or more abnormalities to be present in 96% of incontinent patients (but only in 26% of controls; i.e., 74% specificity), 2 or more abnormalities in 93% (94% specificity), 3 or more in 74% (97% specificity), and 4 or more in 33% (100% specificity). Clinicophysiologic Correlations Four patients had mild (score 0 4), 15 moderate (score 5 8), and 11 severe incontinence (score 9 12). The severity of incontinence, measured by the incontinence scale, was unrelated to the duration of symptoms (R 0.3; P 0.1) or to the age of the patients (R 0.1; P 0.3). The clinical severity of incontinence showed a good inverse correlation with both the force of levator ani contraction (R 0.84; P ) and with the squeeze pressure (R 0.74; P ), and also correlated with the number of physiologic abnormalities detected by the various tests (R 0.50; P 0.005). Measuringthe squeeze pressure as the absolute pressure peak referenced to atmospheric pressure did not improve the correlation with the severity of incontinence (R 0.66; P 0.001). Incontinence was more severe in patients with impaired than with normal levator ani contraction (score vs , respectively; P 0.001) and in patients with impaired than with normal squeeze pressure (score vs , respectively; P 0.001). However, the number of abnormalities was also higher in patients with impaired than with normal levator ani contraction ( vs , respectively; P 0.05) and in patients with impaired than with normal squeeze pressure ( vs , respectively; P 0.05). No other parameter correlated with the clinical incontinence score (Table 2). By multiple linear regression analysis, levator ani contraction was the independent variable with the strongest relation (P 0.001) to the clinical severity of incontinence (Figure 3). The relative gain in proportion of variance accounted for by addingthe squeeze pressure was only 6%. Squeeze pressures correlated with levator ani contraction (R 0.64; P 0.001), and 77% of patients with impaired squeeze pressures also had impaired levator ani contraction. The proportion of variance accounted for by both variables was 76% (square multiple correlation 0.76). Forcingthe squeeze pressure to enter the regression analysis first, 22% of the residual variance was explained by the levator ani contraction. Abnormal cough reflex was unrelated: (1) to the severity of incontinence (score in patients with impaired and in patients with normal reflex; NS), (2) to external sphincter contraction (squeeze pressures 39 8 mm Hgin patients with impaired and mm Hgin those with normal reflex; NS but because of the sample size, a type II error cannot be excluded), (3) to levator ani contraction ( gin patients with impaired and gin those with normal reflex; NS), and (4) to any other physiologic or clinical parameters that were evaluated. Response to Treatment Incontinent patients evaluated 6 months after startingbiofeedback treatment showed a significant improvement in the clinical incontinence score ( score posttreatment vs score pretreatment; P 0.001) (Figure 4). In 84% of the patients the clinical score improved. By multiple linear regression analysis, levator ani contraction before treatment was the only

6 1446 FERNÁNDEZ FRAGA ET AL. GASTROENTEROLOGY Vol. 123, No. 5 Figure 3. Clinicophysiologic correlations. Levator ani contraction was the independent variable with strongest relation to the severity of incontinence. Figure shows regression line ( ), prediction, and 95% confidence intervals. Figure 4. Response to treatment. Both the incontinence score and the force of levator ani markedly improved after biofeedback treatment. *P 0.05 vs. pretreatment. independent variable that predicted the clinical improvement in response to treatment (R 0.53; P 0.01). Of the various physiologic parameters measured, only 2 exhibited a significant improvement after treatment: the force of levator ani contraction and basal anal pressures (Figure 4). Levator ani contraction improved up to a level ( g; P 0.05 vs. pretreatment) that was no longer statistically different from that in healthy or constipated control groups. Overall basal pressure improved significantly (52 5 mmhg; P vs. pretreatment), but only to a modest extent, up to a level significantly lower than in healthy and constipated controls (P 0.05). Squeeze pressure exhibited an inconsistent and not statistically significant increment (up to mm Hg; NS vs. pretreatment) (Figure 4), and still remained well below the level in healthy and constipated controls. The squeeze pressure, measured as the absolute pressure peak referenced to atmospheric pressure, still showed a poorer improvement after treatment (83 9 mm Hgvs mm Hgpretreatment; NS). None of the other parameters, includingrectal perception (threshold for first sensation at 40 4 ml; NS vs. pretreatment), was modified after treatment. In the group of incontinent patients restudied 3 months after the initial evaluation, while on the waiting list for biofeedback treatment (n 12), no changes were detected in the clinical score ( 2% 8% change; NS) or in any of the physiologic parameters measured (levator ani contraction 7% 4% change, basal pressure 15% 8% change, squeeze pressure 2% 6% change, rectal volume of discomfort 3% 8% change; NS for all). In contrast, the incontinent patients that were evaluated 3 months after startingbiofeedback treatment (n 23) already exhibited significant changes in clinical score, basal pressures, and levator ani force (P 0.05 vs. pretreatment for all) that were not significantly different than the effects after 6 months of treatment (n 25). No clinical or physiologic differences in the initial evaluation were found between the groups of patients with and without biofeedback treatment (Table 3).

7 November 2002 LEVATOR ANI AND ANAL CONTINENCE 1447 Table 3. Initial Evaluation of Incontinent Patients Allocated to Groups With and Without Treatment Biofeedback group (n 25) No treatment group (n 12) p b Age (yr) NS Sex, F/M 21/4 12/0 NS Incontinence score NS Basal pressure (mm Hg) NS Squeeze pressure (mm Hg) NS Impaired cough reflex, n (%) 13 (52%) 5 (42%) NS Rectal discomfort (ml) NS Rectal compliance (mm Hg) a NS Levator ani force (g) NS NS, not significant; F, female; M, male. a Rectal pressure at 80 ml intrarectal volume. b No differences were found between both patients groups (P 0.05). Reproducibility Data Levator ani contraction measured on 2 separate days showed a high reproducibility ( gin the first and gin the second measurement; NS) with an intraindividual coefficient of variation of 4.6% 1.1%. Applyingthe Bland-Altman test, 25 in all but one subject, the difference between the 2 measurements lay within 2 SD of the mean difference of the group. Rheologic Properties of Levator Ani and External Anal Sphincter Both intermittent and sustained contractions showed different properties of levator ani and external sphincter, without differences between healthy subjects and constipated controls. Hence, pooled data are provided. After 2-minute, intermittent, brief contractions (repeat squeezes of 2-s duration at 2-s restingintervals), the squeeze pressure produced by the external sphincter decreased significantly (by 11% 4%; P 0.005), but the contraction force of the levator ani did not (by 7% 4%; NS). Duringsustained contraction (Figure 5), the levator ani achieved the maximal amplitude of contraction slower than the external sphincter ( s vs s, respectively; P 0.05). The duration of the sustained contraction was similar in the levator ani and the external sphincter (45 7 s and 37 4 s, respectively), but the levator ani exhibited significantly less fatigability than the external anal sphincter (Figure 5): the force of contraction at the end of the sustained contraction had decreased by 40% 4% for the external sphincter, but only by 27% 3% for the levator ani (P 0.001). In incontinent patients both the levator ani and the external sphincter were weaker than in the control groups, but the differences between both muscles were equally patent. In fact, the fatigability and the duration of the sustained contraction were not significantly different in incontinent patients than in the control groups, and only the time to achieve the maximal amplitude of contraction of the levator ani ( s) and the external sphincter ( s) was significantly longer than in healthy controls (P 0.05 for both). Dissociability of Levator Ani and External Sphincter Function To further prove the distinctive function of the levator ani and the external anal sphincter, we sought patients with dissociation between both muscles. Among the 50 nonselected incontinent patients included in the study (main and ancillary experiments), a dissociation was found in 13 patients, 5 with impaired levator ani force (below 361 g) but normal external sphincter, and 8 with impaired external sphincter squeeze (below 60 mm Hg) but normal levator ani. The latter were older (63 Figure 5. Example of levator ani and external sphincter sustained contraction. The levator ani exhibited significantly less fatigability than the external anal sphincter. Figure corresponds to constipated patient.

8 1448 FERNÁNDEZ FRAGA ET AL. GASTROENTEROLOGY Vol. 123, No. 5 2 years vs years, respectively; P 0.05) and had lower basal pressures (35 4mmHgvs.59 9mm Hg, respectively; P 0.05), but neither the incontinence score nor the number of associated abnormalities differed between them. Amongthe 28 constipated patients with normal continence included in the main and ancillary studies, a dissociation was found in 4 patients, who had impaired levator ani force with normal external sphincter squeeze. These patients produced excessive strainingduringattempted defecation owingto functional outlet obstruction, but had no other associated abnormality. In 5 additional patients, not included in the main or ancillary studies, a dissociation between both muscles was found: 3 incontinent patients with postsurgical sphincteric damage had preserved levator ani function, and 2 constipated patients with functional outlet obstruction and excessive straininghad weak levator ani but normal squeeze pressure. Hence, poolingall these cases together, in 22 patients we showed a dissociation between the levator ani and the external anal sphincter, 11 patients with weak levator ani but normal sphincter, and the other 11 patients with weak sphincter but normal levator ani. Discussion We have demonstrated the role of the levator ani muscle in anal continence by showingthat levator ani contraction is the independent variable with the strongest relation to the severity of incontinence, as well as the strongest predictive factor of the response to treatment. Furthermore, in contrast to other physiologic parameters, a marked and significant levator ani strengthening was associated with clinical improvement in response to biofeedback therapy. Anal incontinence is multiply determined, and no single defect can be said to be either necessary or sufficient to produce incontinence. In our study, the clinical severity was related to the number of associated abnormalities, and this summation of abnormalities may reflect, at least in part, the general deterioration of anorectal mechanisms that takes place in the development of incontinence. 3,12,27 Amongthe various parameters other than the levator ani contraction found to be impaired in incontinence (basal anal pressure, squeeze pressure, rectal perception and compliance, and the cough reflex), none of them except the squeeze pressure correlated with the incontinence score. Squeeze pressure, which reflects the activity of the external anal sphincter, was in turn correlated with levator ani contraction. By multiple regression analysis, the external anal sphincter was less strongly related to incontinence, and this was further substantiated by the lack of significant improvement of the external sphincter that could explain the clinical improvement after treatment. This relative unresponsiveness of the external sphincter to rehabilitation treatment had been previously reported 7,8 and could be explained by an irreversible damage of the most superficial and susceptible perineal muscular component. Alternatively, improvement in the external anal sphincter squeeze pressure may require more extensive biofeedback training. 28 In a recent study, the squeeze pressure significantly improved after biofeedback treatment, but without differences between responders and nonresponders. 9 Abnormal cough reflex, which is driven by the pudendal nerves, 22,23 did not determine the clinical response to therapy and was not improved by rehabilitation. Nevertheless, we acknowledge that other tests potentially indicative of somatic denervation, such as pudendal nerve latency measurements, 12,13 were not performed in our study. Anal basal pressure was also reduced in a large proportion of incontinent patients, but without relation to the clinical severity of incontinence. Previous studies have shown that basal pressures statistically increased after biofeedback treatment, 29,30 an observation also found in our study. The increment was small and probably clinically irrelevant, but nevertheless a logical interpretation for such change is not clear. Basal pressures largely reflect the tonic smooth muscle contraction of the internal anal sphincter, but an additional contribution corresponds to the tonic contraction of the striated muscular structures that surround the anal canal at different levels (i.e., external sphincter caudally and levator ani in the orad part). 15,17,31 It is not clear why biofeedback treatment would alter internal sphincter tone. Alternatively, increased basal pressure could be related to a greater contribution of the striated musculature, thus, maskingan improvement of external sphincter function. However, squeeze pressure measured as the absolute pressure peak from atmospheric pressure, that is, includingbasal pressures, was not significantly improved after treatment. As previously reported, 10,32 we also found that incontinent patients had increased perception of intrarectal volumes but this dysfunction was not related to the severity of incontinence and was not modified by biofeedback treatment. In contrast, in other studies, impaired rectal perception, specifically the threshold for first sensation, is considered an important contributor to fecal incontinence 9,28 and rectal perception improves after biofeedback treatment. 8,9

9 November 2002 LEVATOR ANI AND ANAL CONTINENCE 1449 To evaluate levator ani contraction we used a direct method that measures the traction of the levator ani, the puborectalis sling, and the other fascicles, on an intrarectal device. An important aspect in this system is the 90 angulation between the axis of the anal canal and the dynamometer, and this was carefully accomplished by means of an articulated support fixed to the pelvic bones. Nevertheless, deviations from this angle may not be critical, for instance, 10 deviation, which is a substantial error, would theoretically underestimate the actual traction only by 1.6% (b a cos, where b is the force measured, a is the actual traction, and is the angle of deviation). The reproducibility of the measurements was carefully validated by repeat measurements on the same and different days. A similar approach has been used previously to evaluate anal continence by measuringthe strength required to pull out intrarectal rubber balls or spheres. 31,33 We wish to acknowledge that in the process of reviewingthe literature, we found a 1969 report by Diamant and Harris 33 who measured levator ani strength by a similar approach and obtained values that were in the same range as in the present study. Levator ani contraction was elicited by askingthe patients to squeeze, the same maneuver that produced contraction of the external anal sphincter, and conceivably a global contraction of the pelvic floor musculature. 11,33 However, levator ani performance was distinctly different than that of the external sphincter, both in physiologic and in pathophysiologic conditions. Overall, the levator ani and the external sphincter exhibited a good correlation, but in some patients failure of one muscle coexisted with a normal function of the other. This dissociation may be more frequent than observed because patients with sphincter damage but preserved levator ani may not develop incontinence in the absence of other pathophysiologic factors. Likewise, the rheologic properties of the levator ani were different, showinga slower reaction and significantly less fatigability, probably related to the different tissue structure and innervation. 11,14,34 The severity of incontinence was measured by usinga cumulative score related to the frequency of incontinence for 3 kinds of intraluminal contents: gas, liquids, and solids, because patients with incontinence to liquids are also incontinent to gas, and incontinence to solids is associated with the former 2. The objectivity and sensitivity of the scale were validated by the improvement detected after biofeedback treatment, but not after sham treatment. Furthermore, the incontinence score correlated with the severity of physiologic damage, specifically, with striated muscle weakness and with the number of abnormalities present. In conclusion, our study indicates that the levator ani plays an important role in preservinganal continence. The indications and significance of levator ani testing remain to be established, but still the present data may have relevant clinical implications in the management of anal incontinence. First, prevention of incontinence, a major socioeconomic issue, should strive at preserving the pelvic floor musculature, as well as the anal sphincters. Second, proper evaluation of incontinence may require objective measurements of the various mechanisms that control continence, includingthe levator ani. Third, the levator ani should be considered a target of treatment in incontinence, and should be included as a key physiologic outcome in the evaluation of treatment effectiveness. References 1. Borrie MJ, Davidson HA. Incontinence in institutions: costs and contributing factors. CMAJ 1992;147: Rockwood TH, Church JM, Fleshman JW, Kane RL, Mavrantonis C, Thorson AG, Wexner SD, Bliss D, Lowry AC. Fecal Incontinence Quality of Life Scale: quality of life instrument for patients with fecal incontinence. Dis Colon Rectum2000;43: Whitehead WE, Wald A, Norton NJ. Treatment options for fecal incontinence. Dis Colon Rectum2001;44: Wald A. Biofeedback therapy for fecal incontinence. Ann Intern Med 1981;95: Whitehead WE, Burgio KL, Engel BT. Biofeedback treatment of fecal incontinence in geriatric patients. J AmGeriatr Soc 1985; 33: Enck P. Biofeedback training in disordered defecation. A critical review. Dig Dis Sci 1993;38: Loening-Baucke V. Efficacy of biofeedback training in improving faecal incontinence and anorectal physiologic function. Gut 1990;31: Glia A, Gylin M, Akerlund JE, Lindfors U, Lindberg G. Biofeedback training in patients with fecal incontinence. Dis Colon Rectum 1998;41: Chiarioni G, Bassotti G, Stegagnini S, Vantini I, Whitehead WE. Sensory retraining is key to biofeedback therapy for formed stool fecal incontinence. AmJ Gastroenterol 2002;97: Rasmussen O, Christensen B, Sorensen M, Tetzschner T, Christiansen J. Rectal compliance in the assessment of patients with fecal incontinence. Dis Colon Rectum1990;33: Whitehead WE, Schuster MM. Anorectal physiology and pathophysiology. AmJ Gastroenterol 1987;82: Azpiroz F, Enck P, Whitehead WE. Anorectal functional testing: review of collective experience. AmJ Gastroenterol 2002;97: Diamant NE, Kamm MA, Wald A, Whitehead WE. AGA technical review on anorectal testing techniques. Gastroenterology 1999; 116: Percy JP, Neill ME, Swash M, Parks AG. Electrophysiological study of motor nerve supply of pelvic floor. Lancet 1981;1: Morgan CN. The surgical anatomy of the ischiorectal space. Proc R Soc Med 1949;42: Milligan ETC, Morgan CN. Surgical anatomy of the anal canal with

10 1450 FERNÁNDEZ FRAGA ET AL. GASTROENTEROLOGY Vol. 123, No. 5 special reference to anorectal fistulae. Lancet 1934;II: Strohbehn K. Normal pelvic floor anatomy. Obstet Gynecol Clin North Am1998;25: Whitehead WE, Wald A, Diamant NE, Enck P, Pemberton JH, Rao SS. Functional disorders of the anus and rectum. Gut 1999;45: II55 II Thompson WG, Longstreth GF, Drossman DA, Heaton KW, Irvine EJ, Muller-Lissner SA. Functional bowel disorders and functional abdominal pain. Gut 1999;45:II43 II Schuster MM, Hookman P, Hendrix TR, Mendeloff AI. Simultaneous manometric recording to internal and external anal sphincter reflexes. Bull Johns Hopkins Hosp 1965;116: Burleigh DE. Ng-nitro-L-arginine reduces nonadrenergic, noncholinergic relaxations of human gut. Gastroenterology 1992;102: Parks AG, Porter NH, Melzack J. Experimental study of the reflex mechanism controlling the muscles of the pelvic floor. Dis Colon Rectum1962;5: Sun WM, Read NW, Donnelly TC. Anorectal function in incontinent patients with cerebrospinal disease. Gastroenterology 1990;99: Sun WM, Read NW, Miner PB. Relation between rectal sensation and anal function in normal subjects and patients with faecal incontinence. Gut 1990;31: Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1: Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982; 143: Engel AF, KammMA, BartramCI, Nicholls RJ. Relationship of symptoms in faecal incontinence to specific sphincter abnormalities. Int J Colorectal Dis 1995;10: Heymen S, Jones KR, Ringel Y, Scarlett Y, Whitehead WE. Biofeedback treatment of fecal incontinence. A critical review. Dis Colon Rectum2001;44: Guillemot F, Bouche B, Gower-Rousseau C, Chartier M, Wolschies E, Lamblin MD, Harbonnier E, Cortot A. Biofeedback for the treatment of fecal incontinence. Long-term clinical results. Dis Colon Rectum1995;38: Keck JO, Staniunas RJ, Coller JA, Barrett RC, Oster ME, Schoetz DJ, Roberts PL, Murray JJ, Veidenheimer MC. Biofeedback training is useful in fecal incontinence but disappointing in constipation. Dis Colon Rectum1994;37: Williams AB, CheethamMJ, BartramCI, Halligan S, KammMA, Nicholls RJ, Kmiot WA. Gender differences in the longitudinal pressure profile of the anal canal related to anatomical structure as demonstrated on three-dimensional anal endosonography. Br J Surg 2000;87: Read NW, Harford WV, Schmulen AC, Read MG, Santa AC, Fordtran JS. A clinical study of patients with fecal incontinence and diarrhea. Gastroenterology 1979;76: Diamant NE, Harris LD. Comparison of objective measurement of anal sphincter strength with anal sphincter pressures and levator ani function. Gastroenterology 1969;56: Beersiek F, Parks AG, Swash M. Pathogenesis of ano-rectal incontinence. A histometric study of the anal sphincter musculature. J Neurol Sci 1979;42: Received July 25, Accepted August 8, Address requests for reprints to: Fernando Azpiroz, M.D., Digestive System Research Unit, Hospital General Vall d Hebron, 08035Barcelona, Spain. fernando.azpiroz@wol.es; fax: (34) Supported in part by the Spanish Ministry of Education (Dirección General de Enseñanza Superior del Ministerio de Educación y Cultura, PM ), the National Institutes of Health (grant DK 57064), and the European Commission (grant QLK6-CT ). The authors thank Anna Aparici, Maite Casaus, and Purificación Rodriguez for technical support, and Gloria Santaliestra for secretarial assistance.