Background
Treatment of urgency urinary incontinence has focused on pharmacologically treating detrusor overactivity. Recent recognition that altered perception of internal stimuli (interoception) plays a role in urgency urinary incontinence suggests that exploration of abnormalities of brain function in this disorder could lead to better understanding of urgency incontinence and its treatment.
Objective
We sought to: (1) evaluate the relationship between bladder filling, perceived urgency, and activation at brain sites within the interoceptive network in urgency urinary incontinence; (2) identify coactivation of other brain networks that could affect interoception during bladder filling in urgency incontinence; and (3) demonstrate interaction between these sites prior to bladder filling by evaluating their resting-state connectivity.
Study Design
We performed an observational cohort study using functional magnetic resonance imaging to compare brain function in 53 women with urgency urinary incontinence and 20 controls. Whole-brain voxelwise analyses of covariance were performed to examine differences in functional brain activation between groups during a task consisting of bladder filling, hold (static volume), and withdrawal phases. The task was performed at 3 previously established levels of baseline bladder volume, the highest exceeding strong desire to void volume. All women continuously rated their urge on a 0- to 10-point Likert scale throughout the task and a mixed measures analysis of variance was used to test for differences in urge ratings. Empirically derived regions of interest from analysis of activation during the task were used as seeds for examining group differences in resting-state functional connectivity.
Results
In both urgency urinary incontinent participants and controls, changes in urge ratings were greatest during bladder filling initiated from a high baseline bladder volume and urgency incontinent participants’ rating changes were greater than controls. During this bladder-filling phase urgency incontinent participant’s activation of the interoceptive network was greater than controls, including in the left insula and the anterior and middle cingulate cortex. Urgency incontinent participant’s activation was also greater than controls at sites in the ventral attention network and posterior default mode network. Urgency incontinent participant’s connectivity was greater than controls between a middle cingulate seed point and the dorsal attention network, a “top-down” attentional network. Control connectivity was greater between the midcingulate seed point and the ventral attention network, a “bottom-up” attentional network.
Conclusion
Increasing urge was associated with greater urgency incontinent participant than control activation of the interoceptive network and activation in networks that are determinants of self-awareness (default mode network) and of response to unexpected external stimuli (ventral attention network). Differences in connectivity between interoceptive networks and opposing attentional networks (ventral attention network vs dorsal attention network) were present even before bladder filling (in the resting state). These findings are strong evidence for a central nervous system component of urgency urinary incontinence that could be mediated by brain-directed therapies.
Introduction
Urgency urinary incontinence (UUI), involuntary urine loss associated with urgency, affects millions of women daily and significantly impairs their quality of life. Treatment of UUI with medications has focused on mediating detrusor overactivity and resulting incontinence. Compliance with these medications is limited by frequency of their side effects relative to the frequency of their therapeutic success. Recognition that altered perceptual awareness plays a role in UUI suggests that an alternative approach, addressing abnormalities in brain function, may have a role in UUI treatment.
Women with urgency incontinence manifest abnormal activation of portions of the brain that govern interoception, the perception and interpretation of physiologic stimuli arising within the body. These abnormalities, and the effect they have on other regions in the brain, likely modulate abnormal storage in UUI. Abnormalities in interoception also have the potential to modify urge perception, which may be important in the genesis or persistence of UUI.
Most investigation of abnormal brain activation relies on functional magnetic resonance imaging (fMRI) performed using blood oxygen level–dependent (BOLD) contrast imaging. fMRI can assess localized brain activity that occurs in response to administered stimuli or prompted tasks. Areas of the brain that show coherent neural activation and deactivation are described as demonstrating “functional connectivity,” and together constitute “functional networks.” The definition of functional networks, based on temporal correlations of brain activation, has proven a very robust technique, demonstrating reproducible spatial localization of these networks among many populations.
Each functional network interconnects spatially distinct areas of the brain that together implement a unique component of cognition. Individual networks have been identified as having nodes at consistent anatomical sites. Representative nodes include those in the interoceptive network (anterior cingulate cortex [ACC], insula), dorsal attention network (DAN) (dorsolateral prefrontal cortex [DLPFC] or “frontal eye fields”), the ventral attention network (VAN) (ventrolateral prefrontal cortex [VLPFC] and temporoparietal junction), and the default network (medial temporal lobe, posterior cingulate cortex [PCC]) ( Table 1 ).
| Term | Abbreviation | Description |
|---|---|---|
| Anterior cingulate cortex | ACC | Anterior portion of cingulate gyrus located in medial cerebral hemisphere, important in emotion and pain processing |
| Brodmann area | BA | Cerebral cortical regions mapped based on cell architecture |
| Brain oxygen level dependent | BOLD | fMRI sequence that detects local changes in blood oxygen levels caused by increased cerebral blood (hemodynamic response) at sites of increased neural activity |
| Default mode network | DMN | Brain regions activated when mind is not engaged in cognitive tasks and deactivated during cognitive tasks; nodes include ventromedial PFC and posterior cingulate cortex |
| Dorsal attention network | DAN | Brain regions associated with focused attention on external stimuli responsible for goal-directed “top-down” processing |
| Dorsolateral prefrontal cortex | DLPFC | Region in PFC; component of executive control network and DAN |
| Insula | INS | Cerebral cortical region within lateral sulcus; important component of interoceptive system |
| Middle cingulate cortex | MCC | Middle portion of cingulate gyrus located in medial cerebral hemisphere; activated in emotion and pain processing as well as selection of motor response |
| Posterior medial cortex | PMC | Brain region including precuneus and posterior cingulate; functions as posterior node of DMN |
| Regressor | Hypothesized time course of BOLD activation caused by manipulations of independent variable or by another known source of variability a | |
| Ventral attention network | VAN | Brain regions activated in response to unexpected behavioral stimuli, “bottom-up” network; includes temporoparietal junction and ventral frontal/PFC |
| Ventrolateral prefrontal cortex | VLPFC | Brain region located in PFC; region that is part of VAN |
a Huettel SA, Song AW, McCarthy G, eds. Functional magnetic resonance imaging. Glossary. Sunderland (MA): Sinauer Associates; 2014 G-12.
Pioneering fMRI work by Griffiths et al reported increased activation of the interoceptive network, including the insula and ACC, in response to bladder filling among UUI patients. Within the interoceptive network the insula functions as the key switching center for processing visceral sensation while the cingulate cortex is responsible for integrating emotional context with interoception. Similar abnormalities of interoceptive network activation are seen in patients with fibromyalgia and irritable bowel syndrome (IBS), diseases sometimes termed “hypervigilant” states. In hypervigilant states, interoceptive network activation is heightened and its connectivity to other neural networks is altered.
More complete understanding of how hypervigilant changes in brain function affect urge perception may help guide therapy that is directed toward the brain rather than the bladder. We sought a unique approach that would evaluate the relationship between changing bladder volumes and both interoceptive network activation and perceived urge. We also sought to identify coactivation of executive control networks in the prefrontal cortex (PFC) that could affect perception by interacting with the interoceptive network. Other hypervigilant states demonstrate both altered brain activation in response to stimuli and altered patterns of fluctuations in BOLD signal while the individual is at rest. Those brain regions that demonstrate similar fluctuations in BOLD signal over time during rest are considered as having resting-state functional connectivity. These resting fluctuations in signal within a functional network can be much greater in magnitude than changes in a specific brain region during a task or stimulus and, therefore, can provide much greater signal to noise. Accordingly, after identifying sites in the brain that were activated during increased urge we sought to evaluate resting-state functional connectivity between those specific sites and functional networks. To accomplish this, as far as we know, we compared the largest cohort of UUI participants and comparably aged controls yet published.
We hypothesized that UUI participants would experience increased activation of the interoceptive network in response to bladder filling relative to controls and that this activation would be accompanied by increased urge. We also hypothesized that the pattern of PFC executive control network coactivation with the interoceptive network would differ between UUI participants and controls. Last, we hypothesized that sites of altered interoceptive activity in UUI would also show altered resting-state functional connectivity. Identification of differences in brain activation and functional connectivity in women with UUI would identify aspects of brain function that might be targeted by behavioral and other brain-directed therapies, and provide a means to measure the efficacy of those therapies in treating this often refractory abnormality.
Materials and Methods
Participants
The current study represents a subsample from an ongoing randomized controlled trial comparing hypnotherapy vs pharmacotherapy for treatment of UUI ( NCT01829425 ). Female participants were recruited from an academic urogynecology clinic and from the community at large between March 2013 and May 2015. UUI participants were nonpregnant woman ≥18 years old who: (1) had ≥3 UUI episodes/wk for ≥3 months; (2) were without significant neurologic illness or pelvic organ prolapse beyond the hymen; and (3) had overactive bladder (OAB) Awareness Tool scores ≥8. Both participants with UUI and controls without UUI underwent bedside cystometrics at which time “strong desire to void” volumes, defined as bladder volumes eliciting a “persistent desire to pass urine without the fear of leakage,” were recorded. Controls were women between age 46-80 years without UUI, with OAB Awareness Tool Scores <8, who demonstrated strong desire to void volumes of >200 mL. Potential participants were excluded if they had contraindications to magnetic resonance imaging. The University of New Mexico Institutional Review Board approved the study (no. 09-314) and all participants gave written informed consent.
In all, 56 UUI participants and 23 controls participated in this imaging study. Data for 3 UUI participants and 3 controls were excluded secondary to acquisition issues or excessive head motion (3 times the interquartile range relative to their cohort based on framewise displacement). A total of 53 UUI participants and 20 controls were included in the final functional task analysis.
Clinical assessment
All participants completed a 3-day voiding diary and the OAB Awareness Tool questionnaire, and underwent cystometric testing. The OAB Awareness Tool (or OABv-8) is a validated screening tool for OAB. Demographic and clinical data collection included ethnicity, race, age, parity, surgical history, and body mass index.
fMRI scanner tasks
The fMRI task consisted of infusing the bladder with saline over 9 seconds (infuse phase), maintaining that volume over 19 seconds (hold phase), then withdrawing the same volume over 9 seconds (withdrawal phase). Task timing was maintained with Neurobehavioral Systems Presentation software (Neurobehavioral Systems, Berkeley, California). Specifically, a research nurse was given instructions over headphones along with a computerized timed count for all of the different task intervals (ie, infusing and withdrawing fluids from bladder).
Bladder filling and emptying volumes were determined based on participants’ strong desire volumes ( Supplementary material ). The infuse/hold/withdraw cycle was repeated 6 times at low, medium, and high fill (bladder) volumes ( Supplementary Figure 1 ). Because the task was designed to maximize urge but avoid incontinence, only bladder filling during the high-volume infusion exceeded the strong desire to void volume in the majority of participants. Participants continuously (100-Hz sampling frequency) rated their urinary urge using a nonferrous key-press device positioned directly under their right hand. Urge levels were rated on an 11-point Likert scale (0 = no urge to 10 = severe urge) and were recorded in real time during all phases of the scan (fill, hold, and withdraw). To minimize neuronal activation associated with eye movements, subjects were instructed to maintain visual fixation throughout all trials on the centrally presented cross. Three 2 × 3 (group × phase) mixed measures analyses of variance were used to test for differences in 3 direct measures of urge ratings. These were: (1) the maximum rating; (2) the dynamic range of ratings (maximum–minimum); and (3) the number of changes in subjective urge ratings during each phase of the task.
Resting-state connectivity data collection was performed with the participant’s bladder empty prior to the bladder-filling task. During that time participants passively stared at a white cross for 5 minutes. Three additional UUI and 1 control were motion outliers on resting-state fMRI and were excluded from connectivity analysis.
Magnetic resonance imaging and statistical analysis
T1-weighted echo-planar images were collected on a 3T Siemens Tim Trio scanner (Siemens Medical Solutions, Erlangen, Germany) ( Supplementary material ). Functional maps were calculated using functional MRI of the brain software library and automated functional neuroimaging. Time-series images were first de-spiked, temporally interpolated to correct for slice-time acquisition differences, and spatially registered in 2- and 3-dimensional space to the second EPI image of the first run to minimize effects of head motion. The baseline EPI image (ie, used for motion correction) was then aligned with the native T1 image using an affine transformation, followed by an affine alignment of the native T1 image to Talairach space. These 2 matrices were then concatenated and applied to functional data. Time-series data were subsequently blurred using a 6-mm Gaussian full-width half-maximum filter. Deconvolution was used to generate a single hemodynamic response function for the entire bladder-filling task relative to baseline. Specifically, the hemodynamic response was modelled from the onset of the infuse phase through withdrawal, plus 3 seconds to capture the return to baseline state (total time = 44 seconds or 22 images) ( Supplementary Figure 2 ). A separate urge regressor ( Table 1 ) was calculated by convolution to capture neuronal changes uniquely associated with changes in urge across all phases.
Three whole-brain voxelwise analyses of covariance were performed to examine differences in functional activation across groups for each fill condition. The first factor corresponded to group (UUI vs controls), whereas the second factor corresponded to the number of modelled time points in each phase ( Supplementary material ). Group and the group × time interaction were the effects of interest in this statistical framework. An additional voxelwise t test was performed to detect differences in activation corresponding to the urge regressor. All functional results were corrected for false positives at P < .05 ( P < .005; minimum cluster size = 1472 μL) based on 10,000 Monte-Carlo simulations.
Although analysis of intergroup differences is by necessity complex, our prior work allowed for approximation of the necessary sample size. Results from prior evaluation of fMRI during bladder filling in OAB participants suggested that differences in BOLD signal between UUI participants and controls would yield an effect size of 0.8 (Cohen d). An allocation ratio of participants to controls of 3:1 was chosen to provide adequate samples size for subsequent pre-post treatment evaluation of fMRI in UUI participants, which is the specific aim of the previously described parent study. Assuming an effect size of 0.8 and a participant to control allocation of 3:1, a group of 20 controls and 50 participants would achieve a power of 0.80 for detecting intergroup differences (alpha = .05) in percent signal change at prior regions of interest.
Empirically derived regions of interest from analysis of activation were used as seed points for resting-state functional connectivity analysis examining group differences in functional connectivity. Correlations were calculated between the time course of activation in these regions and the time course of activation in other brain sites. Regions of interest and brain sites showing increased connectivity were assigned to corresponding brain network based previously reported Talairach spatial coordinates (as defined by the brain’s anterior and posterior commissures) or Brodmann areas (defined by brain cytoarchitecture).
Materials and Methods
Participants
The current study represents a subsample from an ongoing randomized controlled trial comparing hypnotherapy vs pharmacotherapy for treatment of UUI ( NCT01829425 ). Female participants were recruited from an academic urogynecology clinic and from the community at large between March 2013 and May 2015. UUI participants were nonpregnant woman ≥18 years old who: (1) had ≥3 UUI episodes/wk for ≥3 months; (2) were without significant neurologic illness or pelvic organ prolapse beyond the hymen; and (3) had overactive bladder (OAB) Awareness Tool scores ≥8. Both participants with UUI and controls without UUI underwent bedside cystometrics at which time “strong desire to void” volumes, defined as bladder volumes eliciting a “persistent desire to pass urine without the fear of leakage,” were recorded. Controls were women between age 46-80 years without UUI, with OAB Awareness Tool Scores <8, who demonstrated strong desire to void volumes of >200 mL. Potential participants were excluded if they had contraindications to magnetic resonance imaging. The University of New Mexico Institutional Review Board approved the study (no. 09-314) and all participants gave written informed consent.
In all, 56 UUI participants and 23 controls participated in this imaging study. Data for 3 UUI participants and 3 controls were excluded secondary to acquisition issues or excessive head motion (3 times the interquartile range relative to their cohort based on framewise displacement). A total of 53 UUI participants and 20 controls were included in the final functional task analysis.
Clinical assessment
All participants completed a 3-day voiding diary and the OAB Awareness Tool questionnaire, and underwent cystometric testing. The OAB Awareness Tool (or OABv-8) is a validated screening tool for OAB. Demographic and clinical data collection included ethnicity, race, age, parity, surgical history, and body mass index.
fMRI scanner tasks
The fMRI task consisted of infusing the bladder with saline over 9 seconds (infuse phase), maintaining that volume over 19 seconds (hold phase), then withdrawing the same volume over 9 seconds (withdrawal phase). Task timing was maintained with Neurobehavioral Systems Presentation software (Neurobehavioral Systems, Berkeley, California). Specifically, a research nurse was given instructions over headphones along with a computerized timed count for all of the different task intervals (ie, infusing and withdrawing fluids from bladder).
Bladder filling and emptying volumes were determined based on participants’ strong desire volumes ( Supplementary material ). The infuse/hold/withdraw cycle was repeated 6 times at low, medium, and high fill (bladder) volumes ( Supplementary Figure 1 ). Because the task was designed to maximize urge but avoid incontinence, only bladder filling during the high-volume infusion exceeded the strong desire to void volume in the majority of participants. Participants continuously (100-Hz sampling frequency) rated their urinary urge using a nonferrous key-press device positioned directly under their right hand. Urge levels were rated on an 11-point Likert scale (0 = no urge to 10 = severe urge) and were recorded in real time during all phases of the scan (fill, hold, and withdraw). To minimize neuronal activation associated with eye movements, subjects were instructed to maintain visual fixation throughout all trials on the centrally presented cross. Three 2 × 3 (group × phase) mixed measures analyses of variance were used to test for differences in 3 direct measures of urge ratings. These were: (1) the maximum rating; (2) the dynamic range of ratings (maximum–minimum); and (3) the number of changes in subjective urge ratings during each phase of the task.
Resting-state connectivity data collection was performed with the participant’s bladder empty prior to the bladder-filling task. During that time participants passively stared at a white cross for 5 minutes. Three additional UUI and 1 control were motion outliers on resting-state fMRI and were excluded from connectivity analysis.
Magnetic resonance imaging and statistical analysis
T1-weighted echo-planar images were collected on a 3T Siemens Tim Trio scanner (Siemens Medical Solutions, Erlangen, Germany) ( Supplementary material ). Functional maps were calculated using functional MRI of the brain software library and automated functional neuroimaging. Time-series images were first de-spiked, temporally interpolated to correct for slice-time acquisition differences, and spatially registered in 2- and 3-dimensional space to the second EPI image of the first run to minimize effects of head motion. The baseline EPI image (ie, used for motion correction) was then aligned with the native T1 image using an affine transformation, followed by an affine alignment of the native T1 image to Talairach space. These 2 matrices were then concatenated and applied to functional data. Time-series data were subsequently blurred using a 6-mm Gaussian full-width half-maximum filter. Deconvolution was used to generate a single hemodynamic response function for the entire bladder-filling task relative to baseline. Specifically, the hemodynamic response was modelled from the onset of the infuse phase through withdrawal, plus 3 seconds to capture the return to baseline state (total time = 44 seconds or 22 images) ( Supplementary Figure 2 ). A separate urge regressor ( Table 1 ) was calculated by convolution to capture neuronal changes uniquely associated with changes in urge across all phases.
Three whole-brain voxelwise analyses of covariance were performed to examine differences in functional activation across groups for each fill condition. The first factor corresponded to group (UUI vs controls), whereas the second factor corresponded to the number of modelled time points in each phase ( Supplementary material ). Group and the group × time interaction were the effects of interest in this statistical framework. An additional voxelwise t test was performed to detect differences in activation corresponding to the urge regressor. All functional results were corrected for false positives at P < .05 ( P < .005; minimum cluster size = 1472 μL) based on 10,000 Monte-Carlo simulations.
Although analysis of intergroup differences is by necessity complex, our prior work allowed for approximation of the necessary sample size. Results from prior evaluation of fMRI during bladder filling in OAB participants suggested that differences in BOLD signal between UUI participants and controls would yield an effect size of 0.8 (Cohen d). An allocation ratio of participants to controls of 3:1 was chosen to provide adequate samples size for subsequent pre-post treatment evaluation of fMRI in UUI participants, which is the specific aim of the previously described parent study. Assuming an effect size of 0.8 and a participant to control allocation of 3:1, a group of 20 controls and 50 participants would achieve a power of 0.80 for detecting intergroup differences (alpha = .05) in percent signal change at prior regions of interest.
Empirically derived regions of interest from analysis of activation were used as seed points for resting-state functional connectivity analysis examining group differences in functional connectivity. Correlations were calculated between the time course of activation in these regions and the time course of activation in other brain sites. Regions of interest and brain sites showing increased connectivity were assigned to corresponding brain network based previously reported Talairach spatial coordinates (as defined by the brain’s anterior and posterior commissures) or Brodmann areas (defined by brain cytoarchitecture).
Results
Clinical measures
By design, UUI and control participant age, parity, and degree of prolapse did not differ ( Table 2 ), nor did they differ by race or ethnicity ( P > .05). Body mass index differed between UUI participants and controls ( P = .02). As expected, voiding diaries, OAB Awareness Tool scores and cystometric findings also differed between groups ( Table 2 ).
| Controls, N = 20 | Patients, N = 53 | P value | |
|---|---|---|---|
| Participant characteristics | |||
| Mean (SD) age, y | 53.2 (5.8) | 55.2 (10.8) | .30 a |
| Mean (SD) body mass index, kg/m 2 | 25.9 (5.9) | 30.1 (7.7) | .02 a |
| Mean (%) parity | |||
| 0 | 3 (15) | 13 (24.5) | .61 b |
| 1 | 2 (10) | 4 (5.5) | |
| 2 | 10 (50) | 16 (30.2) | |
| 3 | 5 (25) | 15 (28.3) | |
| 4 | – | 4 (7.6) | |
| 9 | – | 1 (1.9) | |
| Pelvic Organ Prolapse Quantitation (POP-Q) stage (%) | |||
| 0 | 3 (15) | 8 (15.1) | 1.0 c |
| 1 | 6 (30) | 16 (30.1) | |
| 2 | 11 (55) | 29 (54.8) | |
| Prior surgery (%) | |||
| Stress urinary incontinence surgery | 2 (10) | 9 (17.3) | .07 b |
| Prolapse surgery | 2 (10) | 5 (9.6) | 1.00 b |
| Hysterectomy | 5 (25) | 13 (25) | 1.00 b |
| Cigarette smoker | 2 (10) | 5 (9.4) | 1.00 b |
| Validated questionnaire | |||
| Mean (SD) OAB Awareness Tool score | 1.2 (1.4) | 27.8 (6.4) | <.01 a |
| Cystometrics | |||
| Mean (SD) strong desire volumes, mL | 330.5 (97.4) | 182.0 (81.6) | <.01 a |
| Mean (SD) maximal cystometric capacity, mL | 419.0 (128.1) | 251.4 (102.5) | <.01 a |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
