Altered Periaqueductal Gray Functional Connectivity in Overactive Bladder Patients: A Resting-State Functional Magnetic Resonance Imaging Study Using Clustering Analysis

Article information

Int Neurourol J. 2025;29(3):215-222

Publication date (electronic) : 2025 September 30

doi : https://doi.org/10.5213/inj.2550086.043

Susana Fernández Chadily¹, Naomi R. van Houtum^1,*, Marianne E. van Klaveren^1,*, Armelle Knops¹, Gommert A. van Koeveringe^1,2, Mathijs M. de Rijk^,1,2, Job van den Hurk^1,2,3

¹Department of Urology, Mental Health and Neuroscience Research Institute, Faculty of Health, Medicine, and Life Sciences, Maastricht University, Maastricht, The Netherlands

²Department of Urology, Maastricht University Medical Centre, Maastricht, The Netherlands

³SCANNEXUS Ultra-High Field Imaging Centre, Maastricht, The Netherlands

Corresponding author: Mathijs M. de Rijk Department of Urology, Mental Health and Neuroscience Research Institute, Faculty of Health, Medicine, and Life Sciences, Maastricht University, Universiteitssingel 40, Maastricht 6229 ER, The Netherlands Email: m.derijk@maastrichtuniversity.nl

*Naomi R. van Houtum and Marianne E. van Klaveren contributed equally to this study.

Received 2025 April 1; Accepted 2025 July 28.

Abstract

INTRODUCTION

Lower urinary tract symptoms (LUTS) represent a spectrum of conditions defined by various signs and symptoms affecting the bladder, urethra, and surrounding structures. LUTS are generally classified into 3 main categories: storage, voiding, and postmicturition symptoms. Overactive bladder (OAB) syndrome is a subset of storage LUTS. The International Continence Society defines OAB as urgency, with or without urinary incontinence, typically accompanied by increased daytime frequency and nocturia [1]. It is a chronic syndrome that significantly diminishes health-related quality of life [2]. OAB adversely influences social functioning, sexual well-being, occupational performance, and psychological health [3]. Persistent urgency and frequency can lead to diminished self-confidence, social withdrawal, reduced workplace productivity, and emotional distress [4]. In addition to its personal impact, OAB places a substantial financial burden on patients and healthcare systems [5, 6]. Based on 2 large United States databases, monthly healthcare costs for OAB patients were estimated to be approximately 2.5 times higher than those for matched non-OAB controls with shared comorbidities [5]. Although a wide array of treatments exists—from pelvic floor physiotherapy and pharmacological agents to invasive interventions—finding an effective option is often protracted, and many therapies fail to achieve sustained efficacy [7]. A deeper understanding of OAB pathophysiology is therefore essential for optimizing treatment strategies, enabling more targeted interventions, and ultimately improving patients’ quality of life.

The etiology of OAB remains poorly defined, but most theories suggest that abnormal sensory processing leads to urgency, compelling patients to void at lower bladder volumes and at higher frequencies than healthy individuals [8, 9]. Urgency is defined as a compelling need to urinate that is difficult to defer [1]. These sensations are thought to arise from detrusor overactivity during the storage phase, a phenomenon frequently observed in OAB patients. The detrusor muscle is richly innervated and controlled by both central and peripheral nervous systems [1, 10].

The brain-bladder axis is a complex, hierarchically organized network in which numerous pathways function like switching circuits operating in an all-or-none manner [11]. Urinary storage and elimination depend on coordinated interactions among the lower urinary tract (LUT), peripheral ganglia, spinal cord, and brain. Although many aspects of OAB pathophysiology remain unknown, neuroimaging and neurophysiological studies suggest that functional alterations in brain pathways are central to OAB symptoms [12-15]. The periaqueductal gray (PAG), located in the brainstem, is a critical component of this network, serving as a key relay between higher-order decision-making regions and lower reflexive micturition centers [13]. Functional brain imaging studies have demonstrated PAG activity associated with both storage and voiding functions [16, 17]. In particular, functional magnetic resonance imaging (fMRI) has been widely employed to identify brain regions activated during bladder filling and voluntary control of micturition [18-20]. Our group has previously parcellated the PAG into functionally distinct spatial clusters, observing consistent patterns across individuals [21, 22]. More recently, we identified significant differences in functional correlations between PAG subdivisions across bladder states (empty bladder vs. strong desire to void) when comparing OAB patients with healthy controls (HC) [23]. However, it remains unknown whether PAG functional clusters undergo dynamic reorganization over time in these populations. The present study therefore investigates time-dependent changes in the functional connectivity-based organization of the PAG during bladder filling. We hypothesize that the dynamic evolution of PAG functional organization during bladder filling differs between OAB patients and HC. To test this, we used 7 T fMRI, which allows high spatial resolution analysis of PAG connectivity dynamics.

MATERIALS AND METHODS

MRI Data Acquisition and Preprocessing

We collected 420 T2*-weighted multiband echo planar imaging volumes (mb-EPI sequence; acceleration factor=2, MB-factor =2; TR=1,400 msec, TE=22 msec; voxel size=1.1×1.1×1.1 mm) with 40 slices covering the ROI. Total acquisition time was approximately 10 minutes. A T1-weighted whole-brain anatomical scan was also obtained using an MP2RAGE sequence. Only rs-fMRI data were analyzed; no artificial bladder filling occurred during the scan. The full data collection protocol has been described previously by de Rijk et al. [21]. After scanning, participants were assisted out of the scanner and instructed to void in private, and the voided volume was measured. These postscan volumes were compared with the mean voided volumes at urgency level 2 recorded in participants’ micturition diaries to assess whether the scanning protocol altered perceived bladder capacity (Table 1). Anatomical and functional data were processed using BrainVoyager (Brain Innovation, The Netherlands). The preprocessing pipeline included slice scan time correction, motion correction, temporal high-pass filtering (removing frequencies below 0.01 Hz), and correction for geometric distortions caused by non-zero off-resonance fields. Functional data were coregistered with anatomical data and normalized to Montreal Neurological Institute (MNI) space, a standardized spatial framework for inter-subject comparison. Manual optimization was applied to improve alignment of the brainstem with the MNI template. The PAG was manually segmented in the MNI template to create a binary mask, which was applied to the functional data to extract PAG voxels for subsequent analyses.

Table 1.

Mean postscan voided volumes compared with mean voided volumes at urgency level 2 from the 3-day micturition diary in overactive bladder (OAB) patients and healthy controls

Data and Statistical Analyses

All postprocessing and statistical analyses were performed using custom MATLAB (MathWorks, USA) scripts. The 420-volume rs-fMRI dataset was divided into 6 equal temporal blocks of 70 volumes each to assess temporal changes in PAG functional organization (Fig. 1). Blocks of 70 volumes were chosen to balance temporal resolution and signal stability while enabling assessment of changes across the entire 10-minute scan. For each block, voxel-by-voxel connectivity matrices were generated by computing pairwise correlations between the time series of each PAG voxel. To improve data robustness, the lowest 5% of correlations were excluded, as they likely represented weak or non-significant connections [25]. We then applied a MATLAB implementation of the Louvain module detection algorithm to each 70-volume block to parcellate the PAG, yielding 3 functional clusters per dataset [26]. Each cluster comprised PAG voxels with highly correlated BOLD fluctuations, representing functionally distinct PAG subdivisions [21]. Because the Louvain algorithm is stochastic, we ran 200 iterations per block and selected the parcellation with the highest Q-value (modularity statistic) for further analysis [21]. To quantify temporal stability or reorganization of PAG functional clusters, we calculated correlation coefficients (CCs) between the parcellation maps of each block and the final block (volumes 351–420). Each map represented the spatial distribution of 3 PAG clusters, and CC values indicated the degree of spatial similarity between each block and the final block, thereby providing a quantitative measure of changes in PAG organization over time.

Fig. 1.

Schematic overview of the methodological approach. Functional brain images were acquired from the region of interest (periaqueductal gray, PAG; shown in gray) during a 10-minute resting state functional magnetic resonance imaging scan. The scan consisted of 420 volumes, divided into 6 sequential temporal blocks of 70 volumes each. Each block was parcellated using the Louvain module detection algorithm, and correlations were computed between each parcellation map and the final block’s map (black square; volumes 351–420) to assess temporal changes in PAG functional organization.

To test for statistical differences between groups, we calculated the mean CC per block for each group and compared OAB patients with HC. Permutation testing was used to determine whether observed differences exceeded chance. Group labels (“patient” or “control”) were randomly shuffled 10,000 times, and dissimilarities between permuted groups were computed to generate a null distribution under the assumption of no true difference. The P-value was derived by ranking the observed difference against this distribution. To further assess temporal consistency of PAG organization, we compared the change in CC between the first and last blocks within each group. The same permutation testing approach (10,000 shuffles) was applied, and p-values were calculated by comparing observed differences with the null distribution. Statistical significance was set at P≤0.05 after correction for multiple comparisons.

RESULTS

The rs-fMRI data from 10 participants (6 HC and 4 OAB patients) were analyzed. To evaluate changes in PAG functional organization over time, we calculated CCs by comparing the spatial cluster organization of each of the first 5 temporal blocks (volumes 1–70, 71–140, 141–210, 211–280, and 281–350) with that of the final block (volumes 351–420) for each participant. These comparisons are labeled A, B, C, D, and E, respectively, as summarized in Table 2.

Table 2.

Correlation coefficients for each participant

Fig. 2 illustrates CC values across comparisons A–E, plotted for each participant by group. Green lines represent OAB patients, and red lines represent HC. Across all comparisons, OAB patients consistently showed higher CC values than HC, indicating greater stability in PAG functional organization. We then calculated the mean CC across all participants within each group for comparisons A–E (Fig. 3). Statistical testing revealed that OAB patients demonstrated significantly higher average CC values than HC (P1=0.0143), signifying a meaningful group difference in PAG functional organization. These results are detailed in Table 3, which presents average CC values and standard deviations per temporal block, along with corresponding p-values for group comparisons. Additionally, we compared the change in CC between comparison A and comparison E within each group. OAB patients exhibited a significantly smaller change than HC (P2=0.017), suggesting reduced dynamic reorganization of PAG connectivity over time. This pattern indicates that in OAB patients, the PAG maintains a more rigid functional organization, potentially reflecting altered sensory processing or diminished adaptability to bladder sensations.

Fig. 2.

Correlation coefficients comparing the first 5 temporal blocks with the final block (A–E) for all participants. Red lines represent OAB patients, and green lines represent HC. OAB, overactive bladder; HC, healthy control.

Fig. 3.

Average correlation coefficients (CC) for each group across temporal block (A–E) comparisons. Red lines represent OAB patients, and green lines represent HC. Mean values± standard deviation: HC: A, 0.303±0.111; B, 0.359±0.078; C, 0.407±0.130; D, 0.411±0.143; E, 0.475±0.099. OAB: A, 0.575±0.113; B, 0.579±0.060; C, 0.578±0.099; D, 0.615±0.111; E, 0.580±0.117. Statistical results: group mean CC (P1=0.014) and difference in CC between comparisons A and E (P2=0.017). OAB, overactive bladder; HC, healthy control.

Table 3.

Average correlation coefficients (CCs) and standard deviations for each temporal block in healthy controls (HC) and overactive bladder (OAB) patients

DISCUSSION

This study utilized ultra-high-field fMRI to investigate the functional organization of the PAG during resting state in HC and patients with OAB. Our results demonstrated that OAB patients exhibited significantly higher CC values and greater stability in PAG connectivity patterns over time compared to HC. Specifically, OAB patients maintained a more consistent and stable functional organization of the PAG throughout the entire rs-fMRI scan (approximately 10 minutes; 420 volumes), whereas HC showed lower CC values and greater variability in PAG parcellation maps. This variability in HC may reflect the natural filling of the bladder and its associated sensory processing. By contrast, the absence of dynamic changes in OAB patients suggests an altered sensory processing mechanism, potentially related to hyperresponsiveness to afferent signals or impaired central inhibition of the micturition reflex. Consistent with prior work from our group, these findings further indicate that PAG subdivisions differ between OAB patients and HC depending on bladder state. In earlier studies, we parcellated the PAG across the full 420-volume duration of an rs-fMRI scan under both empty bladder and strong desire to void conditions, applying the Louvain module detection algorithm for connectivity analyses. Those analyses revealed high within-group consistency across bladder states but significant between-group differences [23]. The current study extends that work by showing significant time-dependent differences in PAG connectivity between OAB patients and HC. These results suggest that in OAB, underlying brain organization may be “locked” into a more consistent pattern, potentially reflecting chronic alterations in the central processing of bladder-related signals.

The PAG functions as a central hub for integrating afferent information related to bladder filling and plays an essential role in coordinating the transition between storage and voiding [27, 28]. Consequently, afferent input from the LUT is strongly reflected in PAG activity. One possible explanation is that OAB patients experience elevated afferent processing compared to HC, leading to persistent differences in PAG signaling over time. Green et al. [29] demonstrated that the PAG has a key role in autonomic regulation and the fight-or-flight response. Based on this, OAB patients may experience a hyperarousal or alarm state, with PAG hyperactivation contributing to the increased urge to void. Gill et al. [30] further proposed that in OAB, the set point for voiding in brain regions coordinating LUT function may be disrupted, with sacral neuromodulation acting to restore this balance. These insights support the interpretation that PAG signaling in OAB is altered and that disrupted sensory processing may underlie the observed differences. Consistent with this, our results indicate that PAG activation in OAB patients is elevated at early stages, whereas healthy individuals display a progressive increase in correlations between PAG subdivisions across the scan.

Our findings also align with previous reports of altered connectivity between the PAG and other brain regions in OAB patients. Mehnert et al. [31] observed stronger connectivity between the PAG and the right occipital pole in OAB patients, while HC exhibited stronger connectivity between the PAG and the right thalamus. These results further support the view that OAB involves distinct neural mechanisms affecting bladder control and altered sensory integration.

This study has several limitations that must be acknowledged. First, the sample size was relatively small and limited to female participants, with an age mismatch between the OAB and HC groups. Specifically, the OAB group was considerably older, which may have introduced age-related confounding effects on brain connectivity. Moreover, the absence of urodynamic data limited our ability to confirm the presence of detrusor overactivity and fully characterize the OAB subtype. Together, these factors restrict the generalizability and pathophysiological specificity of our findings. Future studies should include larger, more diverse, and better age-matched cohorts of both sexes, and incorporate objective phenotyping tools to more clearly delineate brain-bladder mechanisms in OAB syndrome.

Second, only the initial 10-minute rs-fMRI scan under empty bladder conditions was analyzed. Consequently, our results reflect a subsensory threshold state and do not capture neural activity associated with later stages of bladder filling or urgency. Another limitation is the lack of formal assessments of bladder sensation during this scan. Although participants reported an empty bladder feeling at scan onset following catheterization and bladder emptying, we cannot definitively link the observed connectivity changes to specific sensory states. Extended scanning protocols capturing different bladder sensations, such as the strong desire to void, along with real-time sensation ratings, would provide a more comprehensive picture of dynamic connectivity changes during bladder filling. It is plausible that in healthy individuals, functional connectivity consistency increases as bladder filling progresses, eventually reaching values similar to those observed in OAB patients, who may already exist in this heightened state. Thus, both groups may ultimately exhibit similar connectivity patterns during strong desire to void. Third, although catheterization enabled a standardized bladder state at scan onset, it may have introduced discomfort, potentially influencing resting-state neural activity. Future studies should consider non-invasive alternatives, such as water loading protocols, to reduce participant discomfort while still achieving controlled bladder filling. These methodological considerations should guide the design and interpretation of future research in this area.

Additionally, these approaches could be applied to evaluate the effects of therapeutic interventions for OAB. They may help determine whether connectivity profiles in responders become more similar to those of HC after treatment, thereby clarifying whether successful therapy normalizes PAG connectivity patterns. To deepen our understanding of the neural basis of micturition, future research should also investigate PAG activity in relation to other brain regions. Examining functional connectivity between the PAG and cortical areas, such as the premotor cortex, could further elucidate the PAG’s role as a relay station across different stages of micturition and highlight the importance of its functional organization.

In summary, this study demonstrated that OAB patients show significantly more stable and consistent functional organization of the PAG during rs-fMRI compared to HC. These differences likely reflect underlying pathophysiological alterations in the neural network responsible for bladder control in OAB. A better understanding of these connectivity profiles could inform the development of more effective treatment strategies and improve patient selection for therapy. Future research should build on these findings to assess treatment efficacy and expand knowledge of the neural basis of OAB syndrome.

Article information Continued