Natural bladder filling alters resting brain function at multiple spatial scales: a proof-of-concept MAPP Network Neuroimaging Study

Neural circuitry regulating urine storage in humans has been largely inferred from fMRI during urodynamic studies driven by catheter infusion of fluid into the bladder. However, urodynamic testing may be confounded by artificially filling the bladder repeatedly at a high rate and examining associated time-locked changes in fMRI signals. Here we describe and test a more ecologically-valid paradigm to study the brain response to bladder filling by (1) filling the bladder naturally with oral water ingestion, (2) examining resting state fMRI (rs-fMRI) which is more natural since it is not linked with a specific stimulus, and (3) relating rs-fMRI measures to self-report (urinary urge) and physiologic measures (voided volume). To establish appropriate controls and analyses for future clinical studies, here we analyze data collected from healthy individuals (N = 62) as part of the Multidisciplinary Approach to the Study of Chronic Pelvic Pain (MAPP) Research Network. Participants orally ingested approximately 350 mL of water, and had a 10 min “fuller bladder” rs-fMRI scan approximately 1 h later. A second 10 min “empty bladder” rs-fMRI scan was conducted immediately following micturition. We examined multiple spatial scales of brain function, including local activity, circuits, and networks. We found changes in brain function distributed across micturition loci (e.g., subregions of the salience, sensorimotor, and default networks) that were significantly related to the stimulus (volume) and response (urinary urge). Based on our results, this paradigm can be applied in the future to study the neurobiological underpinnings of urologic conditions.

Study design. The overall design of the MAPP-II SPS for control participants, which is abbreviated compared to that for the UCPPS cohort, includes screening and enrollment of eligible participants during an inclinic screening/eligibility visit (Week 0) during which participants were asked to provide biological samples, undergo a physical and pelvic exam, complete phenotyping questionnaires, and participate in the bladder filling and neuroimaging protocol (described below), as well as a quantitative sensory testing (QST) protocol. After the Week 0 visit, eligible participants were asked to complete a series of online questionnaires 3 months after the initial visit and asked to come back in 6 months for a follow-up and final study visit. The 6-month visit was identical to the baseline visit. This manuscript only focuses on results from the neuroimaging protocol at baseline.

MAPP-II bladder filling and neuroimaging protocol. The MAPP-II bladder filling and neuroimaging
protocol is shown in Fig. 1. This study used a natural bladder filling paradigm based on oral ingestion of water to create two states of urinary urge in each individual-a "fuller" bladder (FB) state with higher urinary urge and an empty bladder (EB) state with lower urinary urge. At specific time points throughout the procedure described below, the experimenter obtained verbal ratings of pain (0-10 scale) and urge to urinate (0-10 scale). Upon arrival to the neuroimaging session, participants first fully voided their bladder. The volume voided was recorded using a graded specimen collection hat. Following the initial void, participants drank 12 oz (~ 350 mL) of bottled water. The start time, end time, and actual volume of water ingested was recorded. Participants then waited for approximately 40 min before going inside the MRI scanner. During this wait time, participants relaxed, changed into MRI safe attire, and filled out questionnaires. At a fuller-bladder state, participants were positioned in the MRI scanner and were instructed to stay still and awake with their eyes open. At this point, a 10-min resting state fMRI (rs-fMRI) was recorded with a fuller bladder (rs-FB). Following rs-FB, the participant left the scanner and was instructed to void fully into the urine collection hat, and the volume voided was again measured. This void volume was used as a proxy measure for the bladder volume during rs-FB. The participant then entered the scanner again and repeated another rs-fMRI scan in the empty bladder state (rs-EB). Following rs-EB, T1-weighted and Diffusion Tensor Imaging (DTI) scans were acquired. This manuscript will focus on the results from rs-FB and rs-EB, which are the most relevant to natural bladder filling. The neuroimaging protocol described was identical for control participants and the UCPPS cohort, although the UCPPS cohort repeated this full procedure (shown in Fig. 1) at additional time points.
Definition of responders to bladder filling. Measures of urge to urinate before and after the rs-FB and rs-EB scans were averaged to create one urinary urge score associated with each scan. We defined "responders" to the bladder filling protocol primarily by those with higher urinary urge during rs-FB compared to rs-EB. However, we also included a small number of participants as responders if they had greater than 0 urinary urge during rs-FB even if urinary urge did not change between rs-FB and rs-EB, since the brain in these individuals could reasonably be assumed to be processing urge information, and lack of change in urinary urge might be corroborated by lack of change in brain activity. To further clarify, a hypothetical participant could have a rs-FB urge score of x, rs-EB urge score of x, and the resulting change of 0. Despite the change value being 0, this  www.nature.com/scientificreports/ participant would be included as a responder because the bladder provocation elicits an non-zero urge rating of x at rs-FB. "Non-responders" had both no change in urinary urge between rs-FB and rs-EB, and no urinary urge at rs-FB. The difference in urge (i.e. rs-FB minus rs-EB) was used for whole-brain regression analyses in responders. Difference in urge could not be used for whole-brain regression as the values were 0 (see group analysis section below).
MRI scanning. The Neuroimaging Core of the G. Oppenheimer Center for Neurobiology of Stress and Resilience (CNSR) at UCLA operated as the hub for neuroimaging operations in MAPP-II. The scanning parameters for rs-fMRI, T1, and DTI imaging have been described as part of the previous MAPP-I study 16 . UCLA-CNSR took several steps throughout the MAPP-II data acquisition period to ensure multi-site MRI quality. Before start of recruitment, UCLA-CNSR obtained qualification scans of a healthy individual from each site to check scanning parameter compliance, as well as obtaining scans from a "travelling human phantom" (the same individual scanned at all 6 sites with all sequences but without the bladder filling protocol). Throughout the study, scans from all sites were uploaded to the UCLA-CNSR repository where they were promptly reviewed for parameter compliance and image quality.
Here we briefly describe the rs-fMRI and T1 parameters. rs-fMRI scans were acquired with a single shot echo planar imaging (EPI) pulse sequences with conventional rectangular Cartesian sampling. Basic pulse sequence parameters were as follows TR = 2000 ms, TE = 30 ms, Flip angle = 77°, FoV = 220 mm × 220 mm, Resolution = 64 × 64, Phase encode direction = A > P, Slice thickness = 4.0 mm, Slice gap = 0.5 mm, Slice acquisition = ascending (not interleaved), Slices per volume = 34-40 to cover entire brain, Phased array acceleration factor = 2, Bandwidth = maximum to accommodate resolution specifications, Orientation = axial-oblique parallel to the line between the anterior and posterior commissures, Number of volumes = 300 (10 min acquisition).
rs-FB and rs-EB images also underwent fMRIprep preprocessing. Reference images were co-registered with 9 degrees-of-freedom to the T1w reference using boundary-based registration (bbregister, FreeSurfer) 24 . Headmotion realignment was performed using mcflirt (FSL 5.0.9) 25 . Images were warped to MNI152NLin2009cAsym standard space and resampled to 2 × 2 × 2 mm voxel dimension to allow for cross-subject comparison. Framewise Displacement (FD) was calculated using Nipype 26 . Six physiological regressors were extracted for principal component-based noise correction based on anatomical CSF and WM masks computed in native space (aCom-pCor) 27 . Following the fMRIprep minimally preprocessed pipeline, the preproc.nii images were skull-stripped using a dilated MNI mask (fslmaths). Based on recent recommendations 28 , six head motion parameters from mcflirt, aforementioned six aCompCor regressors, and high-pass temporal filtering (0.01 Hz) were done simultaneously using 3dTproject function in AFNI. Finally, 3DBlurToFWHM was used to estimate smoothness of each image followed by iterative smoothing until the images reached a target smoothness of 6 mm FWHM.

Motion assessment. A conservative criterion [3 mm maximum Framewise Displacement (max-FD), as
opposed to the recent recommendation of 5 mm] 29 was used to exclude high motion runs to ensure the results are not confounded by motion artifacts. Out of 124 rs-FB and rs-EB runs collected from participants, 9 runs had a max-FD above 3 mm and were not included for further analysis. For this analysis, each participant required a pair of rs-FB and rs-EB for group analysis.
Assessment of local activity: slow-5 fractional amplitude of low frequency fluctuation (fALFF). We first assessed local activity in discrete brain regions using Amplitude of Low Frequency Fluctuations (ALFF). ALFF is a local measure of spontaneous brain activity, which is computed from the power spectrum of the BOLD signal. Fractional ALFF (fALFF) accounts for physiological confounds and individual differences by examining power in the frequency range of interest compared to power in the total frequency range. Here we focused on slow-5 oscillations (0.01-0.027 Hz), as we have previously identified changes specific to slow-5 fALFF in urologic disease compared to healthy controls 30 , and have correspondingly shown that slow-5 fALFF may predictably respond to different repetitive transcranial magnetic stimulation (rTMS) sequences known to both increase and decrease neural activity neuromodulation of BOLD activity in urologic conditions 31 . In order to generate slow-5 fALFF images, the 3dRSFC AFNI function was used instead of the 3dTproject algorithm above, with a bandpass filter set to 0.01-0.027 Hz. The other steps were kept the same. Slow-5 fALFF images were z-standardized by subtraction of the global image mean and division by the global standard deviation.
Scientific Reports | (2020) 10:19901 | https://doi.org/10.1038/s41598-020-76857-x www.nature.com/scientificreports/ Assessment of circuits: seed connectivity of the periaqueductal gray. Circuits (i.e. communication between brain regions), can be assessed through connectivity of discrete seed regions. Here we chose a seed region in the PAG since this region has been an established hub in the brain networks controlling urine storage and voiding 8 . Seed-to-voxel correlation analysis was used to evaluate whole-brain connectivity maps for the PAG. This PAG seed was defined as a 3 mm radius sphere from a comprehensive meta-analysis of the PAG from previous bladder control studies (centroid MNI coordinates x = 1, y = − 25, z = − 12, Fig. 4A) 32 . Averaged fMRI signal from this PAG seed (fslmeants) was used as a GLM regressor for each individual (fsl_glm) to obtain parameter estimates and associated variances. These parameter estimates were then passed on to group level analysis.

Assessment of networks: default mode, salience, and sensorimotor subsystems. Networks
were assessed with Independent Components Analysis (ICA), a spatiotemporal decomposition method to extract the major networks of the brain. ICA was performed on FSL MELODIC on temporally concatenated rs-FB and rs-EB data, with a dimensionality constraint of 25 Independent Components (ICs). The constraint of 25 ICs was in accordance with previous studies where two resting state runs per individual were concatenated 33,34 .
Here, we focused on three major networks involved in bladder control 8 : salience network (SLN, comprised of the anterior cingulate and the anterior insula, involved in encoding of salient stimuli), default network (DMN, comprised of medial prefrontal and posterior cingulate, involved in interoception), and the sensorimotor network (SMN, comprised of the sensory, motor, and supplementary motor areas, involved in motor and sensory responses). Best fit ICs for the SLN, DMN, and SMN were found through spatial correlation with the Yeo 7-network template 35 . Dual regression was used to explore whole-brain connectivity of each network and resultant connectivity parameter estimate and variance maps for each individual were passed up to group-level analyses.
Group analysis for associations with urinary urge and void volume. For analyses at each spatial scale, the resultant images for each individual (i.e. z-standardized slow-5 fALFF, parameter estimates for PAG connectivity, and parameter estimates for SLN, DMN, and SMN connectivity, respectively) were passed up to group-level difference analyses. The FSL Local Analysis of Mixed Effects 36 (FLAME1 + 2) was used to improve mixed-effects variance estimation. We performed whole-brain linear regression analyses to assess the association between images at each spatial scale and measures of bladder perception (i.e. urinary urge) and bladder stimulus (i.e. void volume) at the time of the scan. All regressions were conducted in responders only, as we did not hypothesize brain activity to be highly modulated in non-responders. Non-responders were not used for further analyses. For these analyses in responders, the rs-FB minus rs-EB parameter estimate difference map was calculated for each subject and input along with regressors of demeaned urinary urge and separately, void volume. Age and sex were included as regressors of no interest. Multiple comparisons correction was conducted using gaussian random field (GRF) cluster threshold (Z > 2.3) and significance at p < 0.05.

Results
Participant demographics and characteristics. Basic MAPP-II control data are presented in Table 1  www.nature.com/scientificreports/ rate (5 of 12). One responder had an rs-FB rating of 1, rs-EB rating of 1, and hence the change was zero; this responder was used for further analysis as evoked urge at rs-FB was a non-zero value.
Effects of natural bladder filling on urinary urge in responders. The MAPP bladder filling protocol produced changes in urinary urge related to bladder volume (Fig. 2). Urinary urge was minimal when first assessed after voiding and drinking water at the beginning of the protocol at the "Post-Drink" time-points ( Fig. 2A). Urinary urge was observed to be rising when assessed approximately 20 min after the water consumption at the "20-min Post-Drink" time point (Fig. 2A). Urinary urge reached a maximum during the rs-FB scan, which was approximately 54 min on average after the water consumption ( Fig. 2A). After voiding, the participants had the rs-EB scan, which was again associated with minimal urinary urge ( Fig. 2A). We also found a significant correlation between the difference in urinary urge between the rs-FB and rs-EB scans and the volume voided after the rs-FB scan, which we interpret as a surrogate for the volume of fluid in the bladder during the rs-FB scan (Fig. 2B).
Increased bladder volume and urinary urge is related to increased local brain activity. This analysis was focused on discovering loci in the brain that encoded stimulus (i.e. voided volume) and resultant perception (i.e. urinary urge). A whole-brain linear regression of the change in fALFF from rs-FB to rs-EB with void volume revealed significant clusters (p < 0.05) that encompassed the subgenual Anterior Cingulate Cortex (sgACC), medial Prefrontal Cortex (mPFC), and lateral Orbitofrontal Cortex (lOFC) (Fig. 3A). In addition, a whole-brain linear regression of the change in fALFF from rs-FB to rs-EB with change in ratings of urinary urge from rs-FB to rs-EB revealed significant clusters (p < 0.05) that encompassed the dorsal Anterior Cingulate Cortex (dACC) and the Supplementary Motor Area (SMA) (Fig. 3B). All these associations were in the negative direction-as urinary urge and voided volume increase, slow-5 fALFF decreases (lower slow-5 fALFF values indicate higher neural activity 31 ). Table 2 summarizes these clusters.
Increased bladder volume and urinary urge is related to increased connectivity of the PAG. Similar to local activity, here we were interested in PAG circuits that encoded stimulus and perception. A whole-brain linear regression of the change in PAG connectivity from rs-FB to rs-EB with void volume revealed significant clusters (p < 0.05) that encompassed the Mid-Cingulate Cortex (MCC), Posterior Cingulate Cortex (PCC), and Dorsolateral Prefrontal Cortex (DLPFC) (Fig. 4B). In addition, a whole-brain linear regression of the change in PAG connectivity from rs-FB to rs-EB with urinary urge revealed significant clusters (p < 0.05) that encompassed the MCC, SMA, right Inferior Frontal Gyrus (IFG), and right anterior Insular Cortex (right aINS) (Fig. 4C). Table 2 summarizes these clusters.  (Fig. 5B). In addition, whole-brain linear regression between voided volume and DMN (Fig. 5C) connectivity revealed a significant cluster (p < 0.05) that encompassed the right Inferior Parietal Lobule (IPL) (Fig. 5D). Moreover, whole-brain linear regression between urinary urge and DMN connectivity revealed a significant cluster (p < 0.05) that encompassed the right Inferior Parietal Sulcus (IPS) and the SMA (Fig. 5E). Furthermore, whole-brain linear regression between void volume and SMN (Fig. 5F) connectivity revealed significant clusters (p < 0.05) that encompassed the ventromedial Prefrontal Cortex (vmPFC)/subgenual Anterior Cingulate Cortex (sgACC), right IPL, right Temporoparietal Junction (TPJ), and Left Dorsal Pons (Fig. 5G). Table 2 summarizes these clusters.

Discussion
In the present study we implemented and demonstrated proof of concept of a natural bladder filling paradigm to assess the brain response across a comprehensive set of spatial scales (from local activity to network communication). This experimental procedure relied on natural diuresis instead of an invasive catheter-based urodynamic testing. This paradigm was successful in evoking a modest sense of urinary urge in a large portion of participants, who were designated as responders. The amount of urine collected upon complete voiding was linearly correlated with the perceived urinary urge (i.e. greater fullness of the bladder translated to greater urinary urge). In past brain-bladder studies, inferences about brain activity were obtained from contrasting different phases of bladder fullness (e.g., water infusion contrasted with post-void) but almost none of these studies have shown any associations between brain activity with participant-reported urinary urge or volume. Our study was able to overcome limitations of previous studies by demonstrating specific links of brain metrics with volume of urine voided (a proxy measure of bladder stimulus) and participant-reported urinary urge (resultant perception). We measured brain function across different spatial scales. Local activity in the dACC and the SMA, as measured with fALFF in the slow-5 band, encoded the level of urinary urge. This was consistent with numerous catheter-based bladder infusion studies where the dACC and SMA are reliably activated during strong desire to void and full bladder 10,11,37,38 . The broader ACC regulates midbrain homeostatic centers 39 and also supplies sympathetic outflow to the bladder 40 whereas the SMA provides motor output 31,41 . The ACC and SMA provides signals that ultimately tightens the urethral sphincter and inhibits contraction of the detrusor muscles, allowing the bladder to be filled 8,40 . At the same time, we found that local activity in the sgACC, lateral OFC, and mPFC encoded the volume of urine voided. The mPFC may interact with arousal centers to play a role in shifting focus from ongoing activities unrelated to micturition to focusing on voiding in an appropriate environment (e.g., such as finding a bathroom) 42 . The OFC drives evaluative and goal-directed behavior 43 and lower activation of the OFC has been linked to poor bladder control 6 . Figure 3. Local brain activity is related to changes in urinary urge and void volume. (A) Whole-brain regression of change in slow-5 fALFF (rs-FB minus rs-EB) with void volume showed significant negative associations in the medial Prefrontal Cortex (mPFC), subgenual Anterior Cingulate Cortex (sgACC), and lateral Orbitofrontal Cortex (OFC), such that increases in volume were associated with decreases in fALFF (i.e. increases in local neural activity) in these regions. (B) Similarly, whole-brain regression of change in slow-5 fALFF (rs-FB minus rs-EB) with change in urinary urge (rs-FB minus rs-EB) showed significant negative associations in the dorsal Anterior Cingulate Cortex (dACC) and Supplementary Motor Area (SMA). such that increases in urinary urge were associated with decreases in fALFF (i.e. increases in local neural activity) in these regions. All results have been corrected for multiple comparisons (p < 0.05).

Scientific Reports
| (2020) 10:19901 | https://doi.org/10.1038/s41598-020-76857-x www.nature.com/scientificreports/ Next, we examined circuits of a major hub related to micturition, the PAG. Neurons in the PAG display both tonic and phasic activity in response to both bladder storage and initiation of micturition 44 . These PAG neurons not only receive ascending input from bladder afferents, but also have rich bidirectional interconnections with higher-order regions, such as the prefrontal and limbic regions, which are essential for providing a wide range of conscious and behavioral responses (e.g., monitoring bladder fullness and voiding at an appropriate time) 2,8,45 . Consistent with the literature on the projections of the PAG 32,46 , our study found that during bladder storage, PAG connectivity to insular, frontal, cingulate, and motor subregions encoded the level of volume and urinary urge. These loci might be encoding different PAG signals-for instance, the DLPFC, being a cardinal node of the cognitive control networks, may be providing descending signals to the PAG 47 to inhibit/not excite downstream micturition centers. At the same time, the SMA might be providing a motor output via the PAG to voluntarily engage the pelvic floor muscles to suppress urge and voiding until an appropriate circumstance arises 31,41 .
Finally, we examined large-scale networks in the brain (i.e. how a synchronized set of brain regions that form a network) interact with other brain regions in order to regulate bladder activity. Namely, we focused on the Default Mode (DMN), Sensorimotor (SMN), and Salience (SLN) networks. Several loci implicated in micturition belong to these networks-the mPFC belongs to the DMN, the SMA belongs to the SMN, and the ACC and the insula belongs to the SLN. We found several within-and cross-network connections that encoded urinary urge and volume. As one notable example, we found that the SLN connectivity to the pINS encoded volume; the pINS is a primary center for interoception where visceral signals from internal organs are processed (also known as visceroceptive cortex) 2,39 .
This natural bladder filling provides an easy-to-implement tool that does not depend on catheterization and hence can be scaled to large sample-size neuroimaging studies. In addition, it can be used to phenotype patients with urologic disorders, such as Urologic Chronic Pelvic Pain (UCPPS), which is a target clinical population to study for the MAPP network. Previous research from the MAPP network has shown that patients with hypersensitivity of the bladder can be classified into different clinical phenotypes 5,48 In particular, patient bladder Table 2. Summary of clusters obtained from whole-brain regression analyses with urinary urge and void. www.nature.com/scientificreports/ phenotypes can manifest as (a) neither painful filling or painful urgency, (b) either painful filling or urgency, and (c) both painful filling and urgency. Strikingly, patients in these categories demonstrate a stepwise worsening of both urologic and non-urologic symptoms, and have sex-specific clinical characteristics. Building on the results reported in this study for healthy individuals, the MAPP research network proposes to examine changes in brain function in both UCPPS compared to controls, as well as among the different clinical phenotypes of UCPPS. Furthermore, previous research from the MAPP network has shown that distributed resting brain connectivity pattern at baseline predicts the longitudinal trajectory of pain symptoms 49 ; this prediction may be refined in future studies by better understanding the brain response to the bladder filling stimulus. While the natural bladder filling paradigm described here has a number of advantages, it does not provide rapid control over bladder volume that is achievable in catheter-based urodynamic studies. Therefore, there are more experimental variables to consider in the natural bladder filling approach. For example, is a fixed volume of fluid provided for consumption, or do participants continue to consume fluid until a target degree of urinary urge is reached? In addition, despite our results showing no evidence for age-volume correlation, future versions of this protocol may want to administer an age-corrected drink volume due to the effects of age on bladder physiology. In this study, we report on a paradigm that was designed for use by the MAPP network to study UCPPS patients, many of whom report significant pain and urgency during natural bladder filling. During the design phase of this study, the 350 mL drink volume was chosen so as not to exceed tolerability for UCPPS patients, based on preliminary work assessing feasibility 4 . In future studies, the brain response to bladder filling will be compared between UCPPS patients and the controls described here. However, the 350 mL drink volume was not implemented with healthy individuals in mind and evidently was not sufficient to evoke a large urge for all healthy participants, hence necessitating a subdivision of responders and non-responders. In addition, the mean evoked urge was only ~ 3 out of 10 for the responder group, which is much lower than many previously published urodynamic studies that drive the bladder to full capacity to maximize urge. Therefore, our study only offers a proof of concept of an ecologically relevant task and needs further validation with various samples and higher drink volume in order to (1) maximize the number of responders, (2) to maximize urinary urge, or (3) to maximize the brain response. Another limitation of our subdivision into responder and non-responder is a lack of direct comparison of whole-brain regressions between the two groups, as the urge values for non-responders were centered around zero. Furthermore, unlike urodynamic studies, bladder capacity was not measured, so the percentage of capacity that volume represents is not known-however, we argue that bladder capacity is not and Right anterior Insula (aINS), such that increases in urinary urge were associated with increases in PAG connectivity to these brain regions. All results have been corrected for multiple comparisons (p < 0.05). www.nature.com/scientificreports/ relevant here as urge is a function of the volume present in the bladder (Fig. 2B). While we acknowledge these limitations of the natural bladder filling approach, this report nonetheless demonstrates that the natural bladder filling paradigm does elicit a robust brain response at multiple spatial scales even in healthy individuals, laying the necessary groundwork for analytic approaches in future studies to examine patients with underlying urologic disorders.