Comorbid bladder pain syndrome and vulvodynia - a cross-sectional analysis of the UNICORN-4 study

Abstract

Background

Bladder Pain Syndrome/Interstitial Cystitis (BPS/IC) and vulvodynia often coexist, exacerbating patient symptoms and complicating the diagnosis and treatment. This study aimed to identify distinct subtypes within a BPS/IC and vulvodynia cohort and evaluate their symptom profiles, psychological characteristics, and sexual function indicators.

Methods

A cross-sectional analysis was conducted on 150 female patients diagnosed with BPS/IC and vulvodynia. The patients completed validated questionnaires assessing bladder symptoms, psychological distress (PHQ-9 and GAD-7), and sexual function (FSFI and FSDS-R). Hierarchical and K-means clustering were used to identify patient subgroups.

Results

Three distinct clusters were identified. Cluster 1 exhibited moderate bladder-specific symptoms and psychological distress. Cluster 2 had severe bladder symptoms and the highest psychological distress. Cluster 3, defined as the vulvodynia-predominant subtype, featured severe vulvodynia, significant psychological distress, and minimal bladder symptoms, aligning with a non-urologic pelvic pain phenotype. Sexual function was significantly impaired across all clusters, with Cluster 3 showing the most severe dysfunction.

Conclusions

This study highlights the heterogeneity within BPS/IC and vulvodynia populations. The identification of a vulvodynia-predominant subtype and non-urologic pelvic pain phenotype emphasizes the need for personalized treatment strategies addressing both physical and psychological factors, particularly sexual dysfunction and psychological distress.

Introduction

Bladder Pain Syndrome/Interstitial Cystitis (BPS/IC) is a chronic condition characterized by pelvic pain, pressure, or discomfort perceived to be related to the urinary bladder, accompanied by at least one other urinary symptom, such as persistent urge to void or frequency [1]. Despite established diagnostic criteria and treatment guidelines, the heterogeneity of BPS/IC symptoms and the presence of comorbidities, such as vulvodynia, present significant challenges in both research and clinical practice [2, 3].

Recent studies have highlighted the complex nature of the BPS/IC and its association with various psychosocial factors. Depression, anxiety, and alexithymia have been found to be prevalent among BPS/IC patients, suggesting a intricate interplay between physical symptoms and psychological distress [4, 5]. Moreover, the impact of BPS/IC on sexual function and quality of life has been increasingly recognized, yet remains inadequately addressed in many clinical settings [6,7,8].

The advent of machine learning techniques has opened new avenues for understanding the heterogeneity of BPS/IC. Unsupervised clustering methods have shown promise in identifying distinct phenotypes within the BPS/IC patient population, potentially paving the way for personalized treatment approaches [4, 9,10,11,12]. However, the specific characteristics of these phenotypes, particularly in relation to comorbid vulvodynia and psychological factors, remain unclear [4, 5, 9].

Furthermore, the assessment of sexual function in patients with BPS/IC, especially in those with comorbid vulvodynia, presents unique challenges. The current standardized questionnaires may not fully capture the complexity of the sexual distress experienced by these patients, highlighting the need for more tailored assessment tools [6,7,8].

This study aimed to address these gaps by employing unsupervised machine learning techniques to identify distinct subgroups within a cohort of patients with BPS/IC, with a focus on the comorbidity of vulvodynia. We sought to characterize these subgroups based on their symptom profiles, psychological characteristics, and indicators of sexual function. Additionally, we aimed to explore the limitations of the current sexual function assessment tools in this patient population.

Methods

Patient selection

This prospective cross-sectional study was approved by the Regional Ethics Committee of Kanagawa Dental University in 2023 (approval number 959 as The UNICORN-4 Study). The study was conducted in accordance with the Declaration of Helsinki and all relevant guidelines and regulations. All the participants provided written informed consent. Participants were recruited from the Yokosuka Urogynecology and Urology Clinic (Yokosuka, Japan) between Aug 1, 2023, and Jul 31, 2024. The study population consisted of patients with pelvic pain, bladder pain, and dysuria due to BPS/IC for at least three months. BPS/IC diagnosis was made by specialists based on a detailed medical history, physical examination, and application of high-sensitivity diagnostic criteria, following guidelines [9, 13].

Patients with Pelvic Pain and Urgency/Frequency (PUF) score ≥ 5 were enrolled [9, 14]. The exclusion criteria included other urological or gynecological diagnoses that could cause urinary symptoms or dysuria, history of invasive treatments, pregnancy, diabetes, neurological or rheumatic diseases, and current smoking.

The control group included 82 age-matched healthy women recruited from routine gynecological checkups at the Yokosuka Urogynecology and Urology Clinic. The selection of control participants was based on a PUF score threshold of ≤ 4, as determined from previous research [9]. A previous study analyzed the distribution of PUF scores in both BPS/IC and control groups using kernel density estimation, demonstrating a clear separation in symptom burden between symptomatic and asymptomatic populations. Based on this data-driven approach, a PUF threshold of ≤ 4 was adopted as an objective criterion for defining asymptomatic controls. To ensure methodological consistency, the exclusion criteria for the control group were aligned with those for the patient group, including a history of urinary tract infections, urological or gynecological malignancies, overactive bladder, chronic pain syndromes such as fibromyalgia or irritable bowel syndrome, pelvic organ prolapse (Stage III or higher), prior invasive treatments for urogenital conditions, diabetes, neurological disorders, or rheumatic diseases. These criteria ensured that the control participants were free of conditions that could confound the symptom comparisons with the BPS/IC group.

All the participants provided written informed consent.

Assessment tools

A comprehensive set of validated questionnaires was administered to patients with the support of female nurses in a private environment. Female psychological nurses conducted additional detailed interviews regarding sexual matters. This approach was chosen to provide a supportive environment for discussion of sensitive topics. The assessment tools included:

Numeric rating Scale-11 (NRS-11) [15]

A single-item, 11-point scale (0 = no pain, 10 = worst possible pain) that measures pain intensity. The NRS-11 is widely used in clinical settings to quantify subjective pain levels.

Interstitial Cystitis Symptom Index (ICSI) and Interstitial Cystitis Problem Index (ICPI) [16]: These questionnaires evaluate the severity of interstitial cystitis symptoms (ICSI: 0–20, with higher scores indicating worse symptoms) and their impact on daily life (ICPI: 0–16, with higher scores indicating greater burden). ICSI and ICPI are commonly used to assess the treatment efficacy in BPS/IC patients.

PUF score [14]

A screening tool designed to identify patients with interstitial cystitis and associated pelvic pain. The PUF score measures symptom burden, including urgency, frequency, and pain severity, with scores ranging from 0 to 35, (higher scores indicating greater symptom burden).

The overactive bladder questionnaire short form (OABq SF) [17] and overactive bladder symptom score (OABSS) [18]

These to assess the presence and severity of overactive bladder symptoms. The OABq SF ranges from 0 to 100 (higher scores indicate greater symptom impact), and the OABSS ranges from 0 to 15 (with higher scores indicating more severe symptoms).

Pelvic floor distress Inventory-20 (PFDI-20) [19]

A questionnaire that quantifies the distress and functional impact of pelvic floor disorders, with a score ranging from 0 to 300 (higher scores indicate worse symptoms). PFDI-20 covers symptoms related to urinary, colorectal, and prolapse conditions.

The Patient Health Questionnaire-9 (PHQ-9) [20] is A nine-item scale used to assess depressive symptoms, with a total score ranging from 0 to 27 (0–4 = minimal depression, 5–9 = mild, 10–14 = moderate, 15–19 = moderately severe, 20–27 = severe depression). The PHQ-9 is a validated screening tool for the detection of clinical depression.

Generalized anxiety Disorder-7 (GAD-7) [21]

A seven-item questionnaire that measures symptoms of generalized anxiety disorder, with scores ranging from 0 to 21 (0–4 = minimal anxiety, 5–9 = mild, 10–14 = moderate, 15–21 = severe anxiety). GAD-7 is widely used in both clinical and research settings.

The Toronto Alexithymia Scale-20 (TAS-20) [22]

A self-report measure that evaluates difficulties in identifying and describing emotions, with a total score range of 20–100 (higher scores indicate greater alexithymia). The TAS-20 includes subscales of emotional awareness and externally oriented thinking.

The Female Sexual Function Index (FSFI) [23] and Female Sexual Distress Scale-Revised (FSDS-R) [24] assess female sexual function (FSFI: 2–36, with lower scores indicating worse function) and the degree of distress related to sexual activity (FSDS-R: 0–52, with higher scores indicating greater distress). They are frequently used in studies on sexual dysfunction in women with chronic pain conditions.

Additional measures for sexual aversion

Each instrument was administered in its entirety, with individual items and total scores being recorded.

To complement validated sexual function questionnaires, a proxy measure for sexual aversion was introduced to address the limitations of FSFI and FSDS-R in assessing distress related to sexual activity avoidance. Patients with FSFI scores of 0 were evaluated using a 10-point scale, with higher scores indicating stronger aversion to sexual activity. This measure aims to capture negative emotional responses among patients experiencing pain-related avoidance or trauma.

Additionally, a proxy measure for sexual aversion was developed and used for patients who reported no sexual activity and had an FSFI score of 0. Given the absence of a standardized questionnaire to assess sexual aversion, a 10-point scale was used to evaluate negative emotional responses to sexual activity, with higher scores indicating stronger aversion. This measure was particularly relevant for patients in Cluster 2, where significant distress related to intimate relationships was observed. Statistical comparisons were performed to explore the differences in the psychological impact between sexually active and inactive patients in Cluster 2.

Clinical procedures

The Vulvodynia swab test was conducted at the 2, 4, 8, and 10 o’clock positions of the vaginal vestibule and urethra, and pain levels were evaluated using a visual analog scale (0–10) [25].

Cystoscopy was performed in all patients under local anesthesia using a flexible cystoscope to evaluate bladder mucosal abnormalities characteristic of BPS/IC [26]. The assessment included the documentation of glomerulation, Hunner’s lesions, and other mucosal changes. The severity of glomerulation was graded according to the European Society for the Study of Interstitial Cystitis (ESSIC) criteria, which classifies BPS/IC based on symptom duration (≥ 6 months) and cystoscopic findings [26].

According to the ESSIC classification, BPS/IC can be categorized into three subtypes:

• 1. BPS/IC with Hunner’s lesions: Presence of visible, inflamed ulcerative lesions in the bladder mucosa.

• 2. BPS/IC without Hunner’s lesions, but with bladder inflammation and no ulcerative lesions, but histopathological examination revealed inflammatory changes.

• 3. BPS/IC without bladder inflammation (normal histopathology), absence of visible lesions, or significant inflammation.

Bladder capacity was measured during the procedure, and biopsies were performed when clinically indicated following standard protocols [13]. This classification provides valuable insights into phenotypic variations in BPS/IC, guiding treatment decisions, and prognosis evaluation.

Clustering analysis and subgroup definitions

Hierarchical clustering based on validated BPS/IC questionnaires was performed to evaluate the patient symptom profiles. The procedure involves distance matrix calculation using standardized clinical indicators and Euclidean distance, followed by hierarchical clustering using Ward’s method [9, 10]. The optimal cluster number was determined using the elbow method, and k-means clustering was applied to assign patients to the identified clusters [9, 10]. To evaluate the robustness of the clustering results, the Silhouette Score and Davies-Bouldin Index were computed, following methodologies used in prior BPS/IC clustering studies [11, 27].

To ensure terminological clarity in clustering classification, this study defined “vulvodynia-predominant subtype” as a subgroup characterized by high vulvodynia symptom scores, severe psychological distress, and relatively mild bladder symptoms, based on prior research [9]. This subtype represents BPS/IC patients in whom extravesical symptoms, particularly vulvodynia, are the predominant complaint rather than bladder-related symptoms.

Additionally, the term “non-urologic pelvic pain phenotype” was used to describe patient groups with dominant pelvic pain symptoms, but without significant bladder involvement. This classification is particularly relevant for patients who experience chronic pelvic pain and psychological distress without severe lower urinary tract symptoms, highlighting the need for a broader, non-bladder-centric clinical perspective for assessing BPS/IC.

Sexual aversion and clustering implications

To comprehensively assess sexual dysfunction, this study incorporated a proxy measure for sexual aversion, recognizing that standardized questionnaires such as the FSFI and FSDS-R fail to capture distress stemming from pain-related avoidance or trauma.

For patients reporting no sexual activity (FSFI = 0), a 10-point scale was used to assess negative emotional responses to sexual activity, with higher scores indicating stronger aversion. This measure was specifically applied to Cluster 2, in which sexual trauma-related distress was hypothesized to be more prevalent. Statistical analyses confirmed that sexual aversion scores differed significantly between sexually active and inactive participants within this cluster.

Future research should further refine and validate this measure to ensure a more accurate evaluation of sexual function in BPS/IC and vulvodynia populations.

These definitions were applied to categorize the patient subgroups in the clustering analysis, ensuring a structured interpretation of symptom heterogeneity in BPS/IC.

Exploratory data analysis and statistical methods

Exploratory data analysis using Principal Component Analysis (PCA) was conducted for high-dimensional dataset visualization [9, 10]. Additionally, t-distributed stochastic neighbor embedding (t-SNE) was applied to enhance cluster interpretability in a lower-dimensional space [12].

Descriptive statistics were used to summarize the baseline characteristics. Continuous variables are presented as mean ± standard deviation and categorical variables as counts (%). Kruskal-Wallis tests assessed statistical differences among clusters for each indicator, including all subscales and individual items of the questionnaires, followed by Tukey-Kramer post-hoc comparisons [9]. One-way Analysis of Covariance (ANCOVA) was used to evaluate the differences in vulvar pain scores between the clusters [9]. Confounding factors such as age and BMI were examined using ANCOVA, as previous studies have demonstrated their potential influence on symptom profiles in BPS/IC and vulvodynia populations [28, 29].

Statistical analyses were performed using Python (SciPy, NumPy, Pandas, Matplotlib, Seaborn, Pingouin, and Scikit-learn). For multiple comparisons, Bonferroni correction was applied to control for Type I errors, ensuring rigorous statistical validation [30]. Statistical significance was set at P < 0.05.

Handling missing data and patient exclusion

To address missing data, this study examined the proportion and distribution of missing values across variables. The FSDS-R showed 22.86% missing data, primarily from patients who did not respond to questions on sexual distress. These missing values were not random, but were systematically related to the absence of sexual activity, suggesting a non-random missingness pattern.

To maintain data integrity and ensure comparability across patient clusters, missing FSDS-R values were imputed using the cluster-wise mean values. This approach preserves within-cluster distributions, while mitigating the potential bias introduced by excluding missing responses. The imputation procedure was validated by confirming that the overall data distribution remained consistent before and after the imputation.

Furthermore, patients with Hunner’s lesions were excluded from the study based on their distinct pathophysiology as defined by the ESSIC classification. These patients exhibited significant differences in inflammatory profiles and treatment responses compared to non-Hunner’s BPS/IC patients, necessitating their exclusion to maintain a homogeneous study population.

Result

Patient characteristics

Between Aug 2023 and the end of Jul 2024, 378 patients with urogenital pain visited Yokosuka Urogynecology and Urology Clinic. A total of 235 patients (62.17%) were excluded, including 43 with urinary tract infections, 5 with diabetic neuropathy, 2 with shingles, 4 with malignant bladder tumors, 16 with ureteral stones, 118 with uterine prolapse, 42 with mesh pain due to previous POP surgery, and 15 with a history of pelvic pain for less than 3 months. The remaining 143 patients (37.83%) were classified into the BPS/IC group because they complained of pelvic pain lasting > 3 months, had a PUF score ≥ 5, and met the American Urological Association’s BPS/IC guidelines. Furthermore, 9 patients with Hunner lesions on cystoscopy were excluded from this study. A total of 134 patients were included in this study. Among the patients who underwent cervical cancer screening, 82 age-matched individuals with PUF scores of ≤ 4 were selected as the control group. The patient characteristics are shown in Table 1.

Table 1 Patient characteristics

Cluster analysis

Expanding on our previous research, this study incorporated additional patient characteristics into the clustering analysis. Hierarchical clustering was employed to analyze patient data, including age and Body Mass Index (BMI), alongside previously validated questionnaire scores.

The resulting dendrogram (Fig. 1a) illustrates the relationships between patients based on this expanded set of variables. The elbow method was used to determine the optimal number of clusters (Fig. 1b) by plotting the sum of squared errors (SSE) against the number of clusters. This analysis, incorporating the new variables, again revealed a significant inflection point in the three clusters.

Fig. 1

Clustering analysis of patient data. (a): Dendrogram for Hierarchical Clustering, x-axis: Patients, y-axis: Euclidean distance (unitless). (b): Elbow Method for Determining Optimal Number of Clusters, x-axis: Number of clusters, y-axis: Sum of squared errors (SSE) (unitless). (c): Principal Component Analysis (PCA) of Patient Data with K-means Clustering, x-axis: Principal Component 1, y-axis: Principal Component 2

Fig. 2

Proportion of vulvodynia and other characteristics in each cluster. (a) NRS-11, x-axis: Cluster, y-axis: NRS-11; (b) ICSI, x-axis: Cluster, y-axis: ICSI; (c) ICPI, x-axis: Cluster, y-axis: ICPI; (d) PUF, x-axis: Cluster, y-axis: PUF; (e) OABq SF, x-axis: Cluster, y-axis: OABq SF; (f) OABSS, x-axis: Cluster, y-axis: OABSS; (g) PFDI20, x-axis: Cluster, y-axis: PFDI20; (h) Vulvodynia swab test, x-axis: Cluster, y-axis: Vulvodynia. Clusters 0, 1, and 2 represented patients with BPS/IC, and Cluster 3 represented the control group

Fig. 3

Psychological and emotional characteristics in each cluster. (a) PHQ-9, x-axis: Cluster, y-axis: PHQ-9; (b) GAD-7, x-axis: Cluster, y-axis: GAD-7; (c) TAS-DDF, x-axis: Cluster, y-axis: TAS-DDF; (d) TAS-DIF, x-axis: Cluster, y-axis: TAS-DIF; (e) TAS-EOT, x-axis: Cluster, y-axis: TAS-EOT; (f) TAS (total), x-axis: Cluster, y-axis: TAS (total). Clusters 0, 1, and 2 represent patients with BPS/IC, and Cluster 3 represents the control group

Fig. 4

Sexual function and distress in each cluster. (a) FSFI Desire, x-axis: Cluster, y-axis: FSFI Desire, (b) FSFI Arousal, x-axis: cluster, y-axis: FSFI Arousal; (c) FSFI Lubrication, x-axis: Cluster, y-axis: FSFI Lubrication; (d) FSFI Orgasm, x-axis: Cluster, y-axis: FSFI Orgasm; (e) FSFI Satisfaction, x-axis: Cluster, y-axis: FSFI Satisfaction; (f) FSFI Pain, x-axis: Cluster, y-axis: FSFI Pain; (g) FSFI Total, x-axis: Cluster, y-axis: FSFI Total; (h) FSDS-R Total, x-axis: Cluster, y-axis: FSDS-R Total. Clusters 0, 1, and 2 represented patients with BPS/IC, and Cluster 3 represented the control group

K-means clustering was applied to segregate the patients into three distinct groups. The results, visualized in Fig. 1c, show the distribution of patients across the first two principal components, reflecting the influence of age and BMI, in addition to symptom profiles.

The complete set of variables used in this enhanced analysis included age, Body Mass Index (BMI), NRS-11, ICSI, ICPI, PUF scores, OABq SF, OABSS, and PFDI-20 (for full details on the questionnaires and references, see our previous publication [9]).

Proportion of vulvodynia and other characteristics in each cluster

Figure 2 shows nine indicators for subjects divided into clusters 0, 1, and 2 and a control group. The indicators were as follows: (a) NRS-11, (b) ICSI, (c)　ICPI, (d) PUF, (e) OABq SF, (f) OABSS, (g) PFDI20, and (h) Vulvodynia swab test.

The Kruskal-Wallis test revealed statistically significant differences for all indicators among the clusters (all P < 0.001).

The distribution of vulvodynia scores is particularly noteworthy. The vulvodynia scores in Cluster 2 (5.84 ± 2.60) were significantly higher than those in Cluster 0 (0.53 ± 1.92) and the control group (1.00 ± 3.00) (P < 0.001). One-way ANOVA confirmed this significant difference (F = 9.94, P = 3.67 × 10^− 6).

Conversely, the OABq SF (8.56 ± 4.83) and OABSS (3.07 ± 1.71) scores in Cluster 2 were comparable to those in the control group (OABq SF: 2.17 ± 3.12, OABSS: 0.70 ± 1.00). Bladder pain assessed by ICSI and ICPI was lower in Cluster 2 than in other clusters (ICSI: Cluster 0:9.96 ± 3.14, Cluster 1:12.26 ± 1.77, Cluster 2:6.05 ± 0.69; ICPI: Cluster 0:10.93 ± 1.32, Cluster 1:12.33 ± 1.06, Cluster 2:3.00 ± 0.65).

The identification of distinct clusters, particularly vulvodynia-predominant Cluster 2, characterized by high vulvodynia scores and mild bladder symptoms, underscores the heterogeneity of the condition and the need for tailored assessment and treatment approaches. Cluster 2 aligns with previously reported “vulvodynia-predominant subtypes” in BPS/IC patients [9], reinforcing the importance of evaluating extravesical symptoms and comorbidities such as vulvodynia.

Additionally, Cluster 2 exhibited a symptom pattern that aligns with the definition of “non-urologic pelvic pain phenotype,” as it was primarily characterized by pelvic pain and psychological distress, with minimal urinary symptoms. This suggests that some BPS/IC patients may present with a phenotype dominated by extravesical pain rather than urological manifestations, necessitating a broader perspective in clinical assessment and management.

Psychological and emotional indicators by cluster

Figure 3 displays six psychological indicators for Clusters 0, 1, 2, and the control group: (a) PHQ-9, (b) GAD-7, (c) TAS-DDF, (d) TAS-DIF, (e) TAS-EOT, and (f) TAS (total).

PHQ-9 scores, indicating depression severity, were notably higher in Cluster 2 (11.28 ± 1.82) than in Cluster 0 (10.07 ± 1.81) and the control group (4.57 ± 3.64) (P < 0.001). Similarly, GAD-7 scores, measuring anxiety levels, showed a significant increase in Cluster 2 (10.56 ± 2.13) compared to Cluster 0 (9.87 ± 1.96) and the control group (4.01 ± 3.22) (P < 0.001).

TAS-EOT (Cluster 2:24.44 ± 2.69, control: 14.26 ± 6.57) and TAS (total) (Cluster 2:114.63 ± 7.63, control: 68.00 ± 27.11) scores in Cluster 2 were comparable to the control group. Emotional clarity, assessed by TAS-DIF and TAS-DDF, was lower in Cluster 2 than in the other clusters (TAS-DIF: Cluster 0:12.50 ± 2.85, Cluster 1:11.52 ± 2.30, Cluster 2:13.31 ± 2.52; TAS-DDF: Cluster 0:19.43 ± 1.46, Cluster 1:18.90 ± 1.63, Cluster 2:19.56 ± 1.46).

Sexual function indicators by cluster

Figure 4 presents eight sexual function indicators for all clusters: (a) FSFI Desire, (b) arousal, (c) lubrication, (d) orifice, (e) satisfaction, (f) pain, (g) total FSFI, and (h) total FSDS-R.

FSFI total scores were significantly lower in Clusters 0 and 2 (both 7.20 ± 0.00) than in the control group (23.94 ± 8.29) (P < 0.001). The FSFI Desire and Arousal scores in Cluster 2 (both 1.20 ± 0.00) were lower than those in the control group (4.10 ± 1.44) (P < 0.001).

FSFI Pain scores were lower in Cluster 2 than in other clusters (Cluster 0:1.20 ± 0.00, Cluster 1:2.48 ± 1.21, Cluster 2:1.20 ± 0.00).

The FSDS-R total scores varied across the clusters. Given that 22.86% of the FSDS-R responses were missing, primarily from patients who did not engage in sexual activity (FSFI = 0), cluster-wise mean imputation was applied to maintain comparability across groups. The final FSDS-R total scores after imputation were as follows: Cluster 0:26.84 ± 22.79; Cluster 1:28.59 ± 16.71; and Cluster 2:26.84 ± 22.79.

In this study, the FSDS-R responses were missing in 22.86% of the dataset, predominantly from patients who did not engage in sexual activity (FSFI = 0). Given this non-random missingness, cluster-wise mean imputation was performed in which missing values were replaced by the mean FSDS-R scores within each cluster. This approach ensured that the overall distribution of FSDS-R scores remained consistent while maintaining comparability across clusters. Sensitivity analysis confirmed that imputation did not significantly alter the variance in the FSDS-R scores.

The imputed FSDS-R values followed the within-cluster distributions: Cluster 0 (0.217), Cluster 1 (2.397), and Cluster 2 (2.244). The appropriateness of this imputation was validated by confirming that the overall data distribution remained consistent before and after the imputation.

Notably, seven patients in Cluster 2 reported no sexual activity, with FSFI scores of 0. These individuals experienced psychological trauma related to their partners’ demands for sexual intercourse, which in some cases led to divorce.

To address the limitations of the existing sexual function assessments, a proxy measure for sexual aversion was introduced. A 10-point scale was used to evaluate negative emotional responses to sexual activity, with higher scores indicating stronger aversion.

This measure revealed that patients in Cluster 2 who reported no sexual activity (FSFI = 0) had significantly higher sexual aversion scores (mean: 8.7 ± 1.2) compared to those who were sexually active within the same cluster (mean: 3.2 ± 1.7, t = 3.976, P = 0.00057).

These results highlight the need to develop new, validated tools to assess sexual dysfunction in BPS/IC and vulvodynia patients, as standardized questionnaires such as the FSFI and FSDS-R fail to capture distress stemming from pain-related avoidance or trauma.

These findings highlight the complex interplay among physical symptoms, psychological distress, and sexual function in patients with BPS/IC and vulvodynia. The identification of distinct clusters, particularly vulvodynia-predominant Cluster 2, underscores the heterogeneity of the condition and the need for tailored assessment and treatment approaches.

Discussion

This study provides a comprehensive analysis of patients with BPS/IC, revealing distinct subgroups with varying symptom profiles, psychological characteristics, and indicators of sexual function. A key feature of this study was its focus on the comorbidity of BPS/IC and vulvodynia, as well as the identification of a vulvodynia-predominant subtype and a non-urologic pelvic pain phenotype. The three clusters identified through hierarchical and k-means clustering offered new insights into the heterogeneity of BPS/IC. Our results were similar to those reported by Okui et al. [9], Mwesigwa et al. [10], and Nettey et al. [27]. All these studies used unsupervised machine learning to identify different subtypes of BPS/IC.

Our analyses revealed three distinct clusters. Cluster 0 showed moderate symptoms across most measures, and represented a “typical” BPS/IC profile. Cluster 1 exhibited the highest severity of bladder-specific and overactive bladder symptoms. Cluster 2 presented a unique profile with severe vulvodynia and psychological distress but relatively milder bladder symptoms, along with the most severe sexual dysfunction [9, 10, 27].

Cluster 2 exhibited severe vulvodynia and psychological distress but relatively mild bladder symptoms, a pattern that aligns with prior definitions of a vulvodynia-predominant subtype in BPS/IC patients [9]. Identification of this subtype reinforces the importance of assessing extravesical symptoms, particularly vulvodynia, when evaluating and treating BPS/IC patients. These findings are consistent with prior research emphasizing the interplay between BPS/IC and genital pain conditions such as vulvodynia [9, 10]. Additionally, Cluster 2 can be classified as a non-urologic pelvic pain phenotype, as it is characterized by chronic pelvic pain and psychological distress without significant urinary symptoms. This distinction is critical in broadening the clinical perspective of BPS/IC, as not all patients present with predominant bladder-related symptoms. Instead, some patients may experience pain localized to extravesical structures, such as the vulvar or pelvic floor muscles, necessitating a multidisciplinary treatment approach [4, 9, 10].

Our study specifically focused on the comorbidity of BPS/IC and vulvodynia, which is a different approach from that of Ricucci et al. [4], despite the use of similar psychological assessments. The high vulvodynia scores and elevated PHQ-9 and GAD-7 scores in Cluster 2 highlight the complex interplay between genital pain, depression, and anxiety in BPS/IC [31]. The analysis of alexithymia scores (TAS-20 and subscales) provided additional insight into the emotional processing difficulties experienced by patients, further emphasizing the role of psychological factors in symptom presentation [32].

The analysis of sexual function revealed significant dysfunction across all clusters, with Cluster 2 showing the most severe impairment. This underscores the profound impact of BPS/IC on intimate relationships and quality of life, extending beyond physical symptoms to psychological and social domains [32, 33]. Given the presence of sexual aversion in some patients, this study introduced a proxy measure for sexual aversion as no standardized questionnaire exists to assess this aspect of sexual dysfunction. This measure allowed for the identification of significant differences in psychological distress between sexually active and inactive patients in Cluster 2. Future research should explore more comprehensive assessment tools that can differentiate among pain-related avoidance, psychological trauma, and generalized distress regarding sexual activity.

While the FSFI and FSDS-R are useful for assessing sexual function, they have limitations in patients with BPS/IC and vulvodynia. The FSFI may underestimate dysfunction in those avoiding sexual activity due to pain, and the FSDS-R does not distinguish between pain-related avoidance and psychological distress. To improve the assessment, future research should refine existing tools or develop a sexual aversion subscale to better capture these aspects.

This study had several limitations that should be acknowledged. First, the single-center recruitment in Japan may limit the generalizability of our findings to other populations. Potential regional, demographic, and cultural differences may influence the symptom presentation and treatment responses. Furthermore, the sample size, which was sufficient for clustering analysis, may limit the statistical power to detect subtle subgroup differences. Future studies should incorporate multicenter recruitment across different geographic regions and diverse populations to validate these findings. Additionally, external validation using publicly available datasets, such as the MAPP Research Network, should be considered in future studies.

Second, as a cross-sectional study, this research could not establish causal relationships between psychological distress, sexual dysfunction, and BPS/IC. Although clustering analysis provides insight into patient subtypes and symptom heterogeneity, it does not allow for conclusions about the temporal sequence of symptom development or causality. Future research should incorporate longitudinal designs to track symptom progression; evaluate treatment responses over time; and determine whether psychological distress or sexual dysfunction precedes, follows, or co-develops with BPS/IC symptoms. Such studies will be critical for refining the diagnostic criteria and optimizing personalized treatment strategies.

Third, a key finding of this study is the limitation of the existing questionnaires in assessing sexual function in patients with comorbid BPS/IC and vulvodynia. Standardized tools such as the FSFI and FSDS-R fail to adequately capture sexual aversion and trauma-related distress, which are significant concerns for many patients. A recent case report [5] highlighted that a patient with comorbid BPS/IC and vulvodynia refused to respond to the FSFI, and FSDS-R did not fully capture their sexual distress. This study implemented a proxy measure for sexual aversion to address this gap, demonstrating a potential alternative for evaluating this aspect of patient experience. However, further validation and refinement of this approach is required.

Despite these limitations, this study provides further evidence supporting the heterogeneity of BPS/IC, identifying a vulvodynia-predominant subtype and non-urologic pelvic pain phenotype. These findings highlight the need for a multidimensional approach for BPS/IC assessment and management that encompasses urological symptoms, genital pain, psychological factors, and sexual function. By recognizing and addressing the complex interplay of symptoms in BPS/IC, particularly its association with vulvodynia and sexual aversion, clinicians can provide more personalized and effective care, potentially improving the treatment outcomes and quality of life for patients with this challenging condition.

Conclusion

This study demonstrated the heterogeneity of BPS/IC and identified three distinct patient subgroups with varying symptom profiles, psychological characteristics, and sexual function, focusing on the comorbidity of BPS/IC and vulvodynia. These findings complement and extend those of previous studies. Our results emphasize the need for a multidimensional approach for BPS/IC assessment and management that encompasses urological symptoms, genital pain, psychological factors, and sexual function. The identification of a vulvodynia-predominant subtype underscores the importance of assessing extravesical symptoms, particularly vulvodynia, in the management of patients with BPS/IC. By recognizing and addressing the complex interplay of symptoms in BPS/IC, especially its association with vulvodynia, clinicians can provide personalized and effective care, potentially improving the outcomes and quality of life of patients with this challenging condition.