Symptom-associated Alterations in Functional Connectivity in Primary and Secondary Provoked Vestibulodynia

INTRODUCTION

Provoked vestibulodynia (PVD) is a chronic vulvar pain disorder characterized by excess sensitivity and severe pain localized to the vulvar vestibule. PVD affects approximately 7 to 16% of women in the reproductive age, and is the leading cause of painful intercourse in this population.[15,46,59] Both central and peripheral alterations in sensory signal processing appear to contribute to PVD pathophysiology[73].

PVD can be subtyped based on the temporal onset of symptoms[34,72]. Women with primary PVD report onset of symptoms at first provoking of vulvar contact/penetration (tampon, sexual intercourse), while those with secondary PVD symptom onset after some period of painless vulvar contact. Women with primary PVD report younger age at symptom onset, longer pain duration, and greater pain during intercourse and are less responsive to existing therapeutic interventions compared to secondary PVD[10,14,16,47,51,90,97]. Overall findings for differences in psychosocial[2,51,89] and sexual functioning[2,90] are inconsistent. Examination of vulvar vestibular biopsy specimens suggests women with primary PVD exhibit greater vestibular nerve fiber density and thickness, neural hypertrophy, vestibular progesterone receptors and less evidence of localized inflammation[35,57,58,95]. Compared to secondary PVD, women with primary PVD have lower sensory detection and pain-pressure thresholds in genital and non-genital areas consistent with central sensitization[21,89,90].

Two studies [43,89] provide tentative support for gray matter and brain activity differences between primary and secondary PVD in default mode[40], salience[32], cognitive control[29,39] and sensorimotor networks involved in processing and modulating ascending sensory signals from the vulva and vagina. Symptom-associated alterations in these networks have previously been reported in studies in heterogenous groups of women with vestibulodynia [11,41–43,71,74,81,91,98] but the influence of onset-subtype on these findings needs to be rigorously examined. Furthermore, the brainstem contains several closely adjacent, interconnected nuclei that play important roles in ascending and descending pain modulation that is relevant to transmission of signals between the brain and the vagina[9,31]. Although understudied, alterations in the connectivity of brainstem nuclei during painful stimulation have been observed in PVD [99]. Ultimately, identification of onset-specific dysregulation of central mechanisms (pain augmentation, dysregulation of endogenous pain modulatory systems, and/or attentional enhancement of pain perception) that may underlie the observed differences in pain profiles and treatment responsiveness by PVD onset subtype are necessary to improve therapeutic approaches.

There is little known about the association between resting state functional connectivity (RSFC) and PVD phenotypes. Symptom-associated alterations in the intrinsic RSFC of the sensorimotor, salience, and default mode networks have been reported in women with PVD compared to healthy controls (HCs)[41]. Also, the global influence of pain processing and modulatory regions (including sensorimotor cortex, thalamus, amygdala, hippocampus, and brainstem) as measured by RSFC strength has been linked to altered steroid hormone biosynthetic and sphingolipid signaling pathways in women with PVD[54,55].

In this study, we performed deep clinical and psychosocial phenotyping and obtained resting state functional magnetic resonance imaging in a large sample of women with PVD and HCs. The primary study aim was to test the hypotheses that the RSFC of the brain and brainstem regions differs between PVD subtypes. Based on prior studies[43,89], we hypothesized onset-specific differences in the connectivity in cognitive control, salience, default mode and somatosensory networks involved in the processing and modulation of sensory signals from the vulva and vagina. Furthermore, we expected PVD onset-dependent connectivity differences in brainstem regions important for ascending and descending pain modulation. As an exploratory aim, we investigated the differences between primary PVD, secondary PVD and HCs on psychosocial assessments, clinical presentation, and symptom measures between the PVD subtypes.

METHODS

Participants

Premenopausal women with PVD were recruited via advertisement on social media, email, UCLA newspaper, and onsite recruiting by study coordinators at the UCLA Center for the Neurobiology of Stress and Resilience and OB/GYN clinics from 2011–2018. PVD was diagnosed by an OB/GYN physician or specialized Nurse Practitioner, both with recognized expertise in examination of women with vulvar pain. The study was registered at ClinicalTrials.gov (NCT02733172).

Inclusion criteria included: 1) Female 18–50 and premenopausal. 2) A minimum three-month history of pain, burning, or irritation, with an intensity of 4/10 or greater localized to the vestibule and provoked by contact of the vestibule. 3) Absence of major medical conditions including neurological, cardiovascular, pulmonary, hepatic, renal, autoimmune, endocrine, or cancer. 4) Willingness to use acceptable contraceptive methods (i.e., barrier, hormonal, or sterilization) if sexually active. 5) Eligible for MRI assessment.

Exclusion criteria included: 1) Presence of major medical, neurological or current psychiatric disorder. 2) Pregnancy or planning a pregnancy, post-partum less than one year, currently breast feeding, or post-menopausal women (defined by no menses for 12 consecutive months). 4) Body mass index (BMI) greater than 35. 6) History of drug abuse or major substance dependence (including nicotine). 7) Current use of centrally acting medication that may interfere with vulva sensory testing. 8) Current use of medications or treatments for vulvodynia initiated less than 3 months prior to the study.

Study Design

PVD patients attended two study visits. The first visit included a medical history interview, physical exam, sensory testing, and completion of questionnaires. The subjects then were scheduled for a second visit for brain imaging. A healthy control age-matched comparison group was acquired from the UCLA Pain and Interception Imaging Network (PAIN) Standardized Repository[56]. All the healthy control women went through the same neuroimaging assessments using the same acquisition protocols and performed on the same scanner at UCLA. Apart from the PVD specific criterion, all inclusionary and exclusionary criteria were the same. By definition healthy controls did not report any chronic pain conditions.

Clinical Assessment of PVD

During the medical history, the onset (primary or secondary) and duration of symptoms as well as presence of other chronic pain disorders was determined. To determine onset subtype, patients were asked if they had pain in vestibule/vaginal opening from time of first contact with a tampon or intercourse (primary) or if they had a period of painless penetration before developing pain ( secondary). Patients with at least one of the 10 chronic overlapping lifetime pain disorders (COPC) described by the Chronic Pain Research Alliance were categorized as having comorbid chronic pain[1].

Vulvar vestibular pain and vaginal muscle tenderness assessment in women with PVD

Vulvar vestibule pain was ascertained using the cotton end of swab to touch the vestibule perpendicularly with enough force to indent the mucosa to a depth of 1/3 of the cotton end for 1 second at 5, 6, 7 (posterior vestibule), 10, 12 and 2 o’clock (anterior/peri-urethral)[13]. Participants were asked to rate the pain at each site on a numeric rating scale (NRS) of 0 to 10 with 0 representing no pain and10 denoting the most severe pain imaginable. The total vulvar vestibular pain score (0–60) was a sum of pain scores across all sites on the cotton swab test. Internal muscle pain, specifically the right and left levator ani muscles (in the vagina) and the perineal complex (at the vaginal entrance), were assessed by placing a single lubricated digit into the vagina then applying approximately 2 kg of pressure for 2 seconds. The examiner’s finger pressure was calibrated before the exam with an algometer. Participants were asked to rate the pain severity each site from 0–10/10. Scores were summed across all locations to compute a total vaginal muscle tenderness score (0–30)[3].

Study Questionnaires

In women with PVD, medical history and PVD associated symptoms and pain were assessed using the Female Sexual Functioning Index (FSFI) [78], Modified Gracely Vulvar Pain Scale [37], and evoked pain/neurophysical pain severity measures[101]. The Complex Medical Symptom Inventory(CMSI) was used to determine the presence of common somatic symptoms of discomfort or pain (e.g. abdominal pain, headache) as well as sensory sensitivity to nonpainful environmental stimuli (e.g. bright light or odors) for the past year[96].

All study participants completed an extensive psychosocial assessment. Participants’ pain experience was further characterized using the Pain Catastrophizing Scale (PCS)[88]. Acute levels of anxiety and depression were evaluated using the Hospital Anxiety and Depression Scale (HADS)[100]. The Spielberger State Trait Anxiety Inventory (STAI)[87]was used to assess trait anxiety level. Extensive trauma history was obtained using the Adverse Childhood Experiences (ACE)[5] and Early Trauma Inventory Self Report-Short Form (ETISR-SF)[17,18] questionnaires. Resilience to stress and pain was measured with the Brief Resilience Scale (BRS)[83]. The Perceived Stress Scale (PSS)[25] was implemented to gauge the subject’s perceived stress.

Brain Imaging

Structural and Functional Imaging Acquisition

All brain imaging data were acquired using a 3.0T MRI scanner (Siemens Trio; Siemens, Erlangen, Germany). High resolution structural brain images were acquired using a magnetization-prepared rapid acquisition gradient-echo (MPRAGE) sequence with the following parameters: repetition time = 2200ms, echo time = 3.26ms, structural acquisition time =5m 12s, slice thickness = 1mm, 176 slices, 256*256 voxel matrix, 1mm voxel size. Resting-state functional magnetic resonance imaging (rs-fMRI) data was acquired using the following parameters: 40-slice whole brain volumes, slice thickness = 4mm, repetition time = 2000ms, echo time= 28ms, resting acquisition time = 10m6s, flip angle = 77°, field of view = 220, 2×2×2 mm voxel size. During the rs-fMRI scan, subjects were instructed to rest with their eyes closed. Neither heart rate nor respiration data was collected during the imaging.

Imaging Analysis

Pre-processing for quality control was performed in Statistical Parametric Mapping 12 (Welcome Department of Cognitive Neurology, London, UK) and included bias-field correction, co-registration, motion correction, spatial normalization, tissue segmentation, and Fourier transformation. Structural images were included based on compliance with acquisition protocol, full brain coverage, minimal motion, Gibbs ringing, absence of flow/zipper and minor atrophy/vascular degeneration. Functional images were included based on compliance with acquisition protocol, full brain coverage, motion estimate of <2 mm in the three directions of translation and three directions of rotation, and <.25mm average framewise displacement[70].

Structural and functional images were entered in CONN 17 toolbox in MATLAB [60] for preprocessing, denoising and analysis. Preprocessing was performed using the default pipeline and included: functional realignment and unwarping, slice-timing correction, structural segmentation & normalization, functional normalization, outlier detection, and 8mm smoothing. Preprocessing output images included: skull-stripped normalized structural volume, gray/white/CSF normalized masks, realigned slice time corrected normalized smoothed functional volumes, subject realignment, and scrubbing first level covariates. Next, we applied linear regression and band-pass filtering simultaneously to remove unwanted motion and physiological effects from the BOLD signal before computing connectivity measures. We removed the effects of white matter and cerebral spinal fluid (5 dimensions each), outliers, linear trends, realignment, and motion (12 regressors: 6 motion parameters + 6 first-order temporal derivatives). A band-pass filter (0.008Hz<f < 0.08Hz) was applied to reduce low and high frequency noise[70]. For each subject, functional connectivity was computed between all brain regions using bivariate correlations and reflects the association between average BOLD time series signals across all voxels in each brain region. RSFC indexes the temporal correlation of spontaneous brain activity during rest among regions comprising distinct functional networks[84]. Importantly, correlational based RSFC provides information on the statistical dependencies between spontaneous activity in two regions across time but is blind to directionality of information between two regions.

Three brain imaging atlases were employed for region of interest (ROI) selection and functional brain network construction. The Schaefer 400 functional cortical parcellation atlas was used to define cortical regions and networks[79]. The Schaefer 400 atlas parcels the brain into 400 functionally meaningful brain regions that comprise 17 networks. These networks include the Somatomotor, Dorsal Attention, Salience/Ventral Attention, Control, Default Mode, Visual, Temporal Parietal, and Limbic (see Figure 1). The temporal parietal network is comprised only by temporal and parietal regions. The Limbic network is comprised by areas in orbital frontal cortex and the ventral temporal pole. The Harvard-Oxford subcortical atlas was used to define subcortical regions including the bilateral thalamus, amygdala, hippocampus, and basal ganglia regions (caudate nucleus, pallidum (global pallidus), putamen, and the nucleus accumbens)[28]. Lastly, the Harvard Ascending Arousal Network atlas was used to define brainstem nuclei including locus coeruleus, mesencephalic reticular formation, parabrachial complex, nucleus reticularis pontis oralis, and the pedunculopontine nucleus bilaterally as well as the dorsal raphe, median raphe, periaqueductal gray, and the ventral tegmental area[30]. Here, we employ an atlas-based investigation of individual brainstem nuclei rather than relying on one region of interest for the entire brainstem. The approach has been successfully employed by others to map structural and functional connectivity of brainstem nuclei [12,77,85,86].

Figure 1.

The Schaefer 400 functional cortical parcellation atlas [79]. The atlas parcels the brain into 400 brain regions that comprise the Somatomotor (SomMot), Dorsal Attention (DorsAttn), Salience/Ventral Attention (SalVentAttn), Control (Cont), Default Mode (Default), Visual, Temporal Parietal(TempPar), and Limbic networks.

Statistical Analysis

Linear contrast analysis within the framework of the general linear model (GLM) was applied within the CONN 17 toolbox to test for group differences in functional connectivity. This approach first defines a model, estimates its parameters, then performs hypotheses testing. After parameter estimation, standard Likelihood Ratio Tests are used to specify and test hypothesis in the form of contrast. Contrasts are evaluated using Wilks’ Lambda statistic and a reported as statistic that approximates a Student’s t-distribution [68]. In addition to group as an explanatory factor, vulvar pain duration and presence of other chronic overlapping pain conditions were entered as covariates in the design matrix when testing the between-subjects contrast comparing primary and secondary PVD. The following between-subjects contrasts were specified to test the hypotheses: 1) Primary PVD-HCs, 2) Secondary PVD-HCs, and 3) Primary PVD-Secondary PVD. Significance was considered at a seed-level corrected false discovery rate (FDR) of 1% (q<.01) for cortical regions and 5% (q<.05) for the smaller brainstem nuclei. The q-value is a p-value that has been FDR adjusted. This multiple-comparison approach corrects for multiple target ROIs, but separately for each seed ROI.

Group differences in demographic, psychosocial and symptom measures were assessed using the independent t-tests in SPSS software version 22.0 (SPSS Inc.) applying a threshold for significance at p<.05. To elucidate significant mean differences, we report the absolute value of the standardized mean change, Cohen’s effect size d. As a rule of thumb, effect sizes are interpreted as small (d=.20), medium (d=.50), or large (d=.80).

An exploratory hypothesis generating analyses correlational analysis was performed to provide insights into the link between the observed differences in functional connectivity between primary PVD and secondary PVD and the psychosocial and pain assessments that differed between primary PVD and secondary PVD at p <.10 (i.e., early life sexual abuse as measured by the ACE and the ETI, ACE household mental abuse, pain catastrophizing, somatic symptoms of discomfort/pain and sensory sensitivity to nonpainful environmental stimuli (CMSI), 24-hour symptom unpleasantness and intensity, and sexual satisfaction (FSFI), see Table 1 and Table 2. HCs were only included in the correlational analyses where they completed assessments (i.e., ACE, ETI). Subject-level pairwise connectivity was extracted in the form of i.e., z transformed correlations. Prior to computing Pearson’s correlations, the connectivity data was residualized for vulvar pain duration and presence of other chronic overlapping pain conditions. A threshold of p<.05 uncorrected was applied for reporting correlations.

Table 1.

Clinical presentation in primary and secondary provoked vestibulodynia

	Primary PVD	Secondary PVD	Comparison of secondary to primary PVD
Female Sexual Functioning Index	13.56(5.97) N = 35	12.17 (6.71) N = 40	t = −0.96, p = 0.34
Desire [0–5]	2.39(0.94) N =40	2.40 (1.11) N = 45	t =0.06, p = 0.96
Arousal [0–5]	2.56(1.50) N = 39	2.10 (1.52) N = 44	t = −1.43, p = 0.16
Lubrication [0–5]	2.76(1.70) N = 39	2.55 (1.87) N = 44	t = −0.52, p = 0.60
Orgasm [0–5]	2.55(1.70) N = 38	2.89 (1.76) N = 45	t = −0.69, p = 0.49
Satisfaction [0–5]	2.67(1.34) N = 39	1.74 (1.19) N = 41	t = −3.24, p = 0.002
Pain [0–5]	1.03(0.98) N = 39	0.96 (1.06) N = 45	t = −0.32, p = 0.75
Vulvodynia Symptom Unpleasantness Rating Past 24 hours [0–20]	2.79(3.61) N = 46	4.95 (4.43) N = 68	t = 7.74, p = 0.007
Vulvodynia Symptom Intensity Rating Past 24 hours [0–20]	3.45(4.58) N = 46	5.18 (4.96) N = 68	t = 1.88, p = 0.058
Vestibular Pain Total (cotton swab)[0–60]	29.70(9.36) N = 45	28.37 (11.02) N = 67	t = −0.66, p = 0.51
Vaginal Muscle Pain Total[0–30]	11.26(7.62) N =43	11.22(7.40) N =60	t = 0.03, p =0.98
Pain Duration (Years)	8.2(7.8) N =46	5.4(5.6) N =68	t = −2.07, p = 0.04
			t = 0.84, p = 0.40
Complex Medical Symptoms Inventory (12 m) [0–35]	4.41(4.3) N=46	6.15 (6.20) n = 66	t = −1.75, p = .08
ComorbidChronicPain * Condition (Lifetime)	30(65.25)   N = 46	45(66.2%)   N = 68	p=1.00

Statistics are presented as mean (standard deviation) or frequency (percentage %)

N = total number of subjects completing the assessment.

^*Fisher’s Exact test.

Abbreviations: PVD=provoked vestibulodynia

Table 2.

Psychosocial characteristic of the sample.

	Healthy Controls	Primary PVD	Secondary PVD	Comparison of primary PVD to HC	Comparison of secondary PVD to HC	Comparison of secondary PVD to primary PVD
Depression & Anxiety
Depression (HAD)[0–21]	1.33(1.90) N = 94	2.56(2.52) N = 46	3.24 (3.26) N = 68	t =2.94, p = 0.004	t =4.33, p =2.0E-6	t = 1.18, p = 0.24
Anxiety (HAD)[0–21]	3.85(3.41) N = 94	6.35(3.65) N = 46	7.12 (3.83) N = 68	t =3.98, p =0.0001	t =5.71, p =5.3E-8	t = 1.07, p = 0.29
Trait anxiety (STAI-T)[20–80]	45.22(9.40) N = 92	54.13(10.52)   N = 45	55.26 (10.95) N = 68	t =5.01, p =2.0E-6	t =6.23, p =4.1E-9	t = 0.55, p = 0.59
Perceived Stress Scale[0–40]	11.26(6.21) N =92	14.74(6.17) N = 46	17.15 (6.81) N = 67	t =3.11, p = 0.002	t =5.57, p=6.7E-8	t = 1.92, p = 0.058
Pain Catastrophizing Scale (PCS)
Rumination[0–16]	3.63(4.17) N =32	5.18(4.14) N =45	6.12 (4.02) N =67	t = 1.62,p = 0.11	t =2.85, p = 0.005	t = −1.20, p = 0.23
Magnification[0–12]	1.59(1.93) N =32	2.53(2.36) N =45	3.35 (2.68) N =66	t =1.85, p = 0.07	t =3.31, p =3.0E-4	t = −1.65, p = 0.10
Helplessness[0–24]	3.59(4.86) N =32	5.71(4.66) N =45	8.06 (5.45) N =67	t =1.93, p = 0.06	t =3.94, p =1.5E-4	t = −2.37, p =0.02
Overall [0–52]	8.81(10.28) N =32	13.42(9.98) N = 44	17.52 (10.92) N = 67	t =1.97, p = 0.05	t =3.78, p=2.7E-4	t = 2.01, p = 0.046
Trauma
Trauma –ACE_Score [Present/Not Present]	1.16 (1.51) N = 89	1.00 (1.12) N =45	1.37 (1.40) N =67	t = −.53, p = .60	t =.91, p = .36	t = 1.47, p = 0.14
Emotional Abuse	10(11.2%) N = 89	4 (8.9%) N =45	8 (11.9%) N =67	p = 0.77	p= 1.00	p= 0.76
Physical Abuse	2(2.2%) N = 89	0 (0%) N = 45	1 (1.5%) N = 67	p = 0.55	p= 1.00	p =1.00
Sexual Abuse	14 (15.7%) N = 89	1(2.2%) N = 45	12 (17.9%) N = 67	p = 0.02	p= 0.83	p = 0.01
Substance Abuse	19(21.3%) N = 89	6(13.3%) N = 45	19(28.4%) N = 67	p = 0.35	p= 0.35	p =0.07
Parental Divorce/Sep	28(31.5%) N = 89	13(28.9%) N = 45	25(37.3%) N = 67	p =.84	p= 0.50	p = 0.42
Household Mental Illness	18 (20.2%) N = 89	19(46.3%) N = 45	22(32.8%) N = 67	p = 0.01	p= 0.10	p = 0.32
Incarcerated Fam Member	3 (2.2%) N = 89	0(0%) N = 45	2(1.8%) N = 67	p = 0.55	p= 1.00	p = 0.52
Violent Parents	9(10.1%) N =89	3 (6.7%) N = 44	3(4.5%) N = 67	p = 0.75	p= 0.24	p = 0.68
Trauma – Early Trauma Inventory
ETI General Score [0–11]	1.59(1.59) N =90	1.38(1.37) N =45	1.53 (1.30) N =66	t = −0.76, p = 0.45	t = −0.24, p = 0.80	t = 0.59, p = 0.56
ETI Physical Score [0–5]	0.99 (1.32) N =90	0.72(0.94) N = 46	0.46 (0.90) N =65	t = −1.39, p =0.17	t = −2.95, p =0.004	t = −1.45, p =0.15
ETI EmotionalScore[0–5]	0.60(1.36) N =90	0.50(1.13) N =46	0.83 (1.36) N =65	t = −0.43, p = 0.67	t = 1.04, p = 0.30	t = 1.35, p =0.18
ETI SexualScore[0–6]	0.36 (0.84) N =90	0.17(0.64) N =46	0.49 (1.03) N =65	t = −1.40, p =0.16	t =0.91, p = 0.36	t = 1.85, p =0.048
ETI Total Score [0–27]	3.53 (3.60) N = 90	2.80 (3.30) N = 45	3.29 (3.10) N = 65	t = −1.18, p = 0.24	t = −0.44, p = 0.66	t = 0.80, p = 0.43
Resilience
Brief Resilience Scale[6–30]	23.09(4.90) N = 70	20.40(4.90) N = 42	20.25 (4.42) N = 60	t = −3.01, p = 0.003	t = −3.68, p =3.4E-4	t = −0.16,p = 0.87
Connor-Davidson Resilience Scale[0–100]	77.34(11.25) N = 71	72.99(12.49) N = 42	72.35 (14.48) N = 60	t = −1.90, p = 0.06	t = −2.22, p = 0.03	t = −0.24,p = 0.82
Persistence[0–32]	25.69(4.49) N =71	24.14(4.89) N =42	24.38 (5.33) N =60	t = −1.71, p = 0.09	t = −1.53, p = 0.13	t = 0.22, p = 0.82
Emotional Cognitive[0–28]	19.94(3.80) N = 71	19.14 (3.99) N = 42	19.07 (4.38) N = 60	t = −1.06, p = 0.29	t = −1.21, p = 0.23	t = −0.08, p = 0.843
Adaptability[0–20]	16.99(2.56) N = 71	16.00(2.69) N = 42	16.08 (2.68) N = 60	t = −1.94, p = 0.06	t = −1.97, p = 0.051	t = 0.15, p = 0.88
Control/Meaning[0–12]	9.63(1.78) N = 71	8.88(2.17) N = 42	8.72 (2.24) N = 60	t = −2.00, p = 0.047	t = −2.61,p = 0.01	t = −0.37, p = 0.71
Meaning[0–8]	5.08(2.44) N = 71	4.88(5.08) N = 42	4.24 (2.44) N = 58	t = −0.44, p = 0.65	t = −1.96, p = 0.053	t = −1.31, p = 0.19

Statistics are presented as mean (standard deviation) or frequency (percentage %)

N = total number of subjects completing the assessment

RESULTS

Subject Characteristics

The sample consisted of 46 primary PVD, 68 secondary PVD and 94 HCs. Participant did not differ with respect age to (M(SD) primary PVD: 28.67yrs (6.68), secondary PVD: 28.85 (7.27) and HCs 29.82 (10.44).

Clinical presentation and symptoms in primary PVD and secondary PVD

Primary PVD reported experiencing chronic vulvar pain for a significantly longer duration than secondary PVD (Table 1; Cohen’ d =0.42). Using the Gracely Vulvar Pain Scale, secondary PVD reported significantly greater vulvodynia symptom unpleasantness in the past 24 hours (d=0.52), with a trend for greater intensity (d=0.36) than patients with primary PVD. No differences in overall levels of sexual functioning as measured by the FSFI were observed between the PVD onset subtypes. However, women with secondary PVD exhibited significantly lower global sexual satisfaction than primary PVD, as measured by the FSFI (Cohen’ d =0.73). No significant difference was observed between primary PVD and secondary PVD in the presence of self-reported comorbid chronic pain conditions. However, women with secondary PVD reported a trend for elevated levels of somatic symptoms of discomfort/pain and sensory sensitivity to nonpainful environmental stimuli (d=0.32). No significant differences on examination were observed between the PVD subtypes in evoked vulvar vestibular pain or vaginal muscle tenderness.

Differences among primary PVD, PVD 2, and HCs on psychosocial functioning

Women with primary PVD and secondary PVD exhibited significantly greater depression (Table 2; d_{primary PVDvHC}=0.58, d_{secondary PVDvHC}=0.75) and state anxiety (d_{primary PVDvHC}=0.72, d_{secondary PVDvHC}=0.91) than HCs, as measured by the HADS. Women with both PVD subtypes also exhibited significantly greater trait anxiety compared to HCs (d_{primary PVDvHC}=0.91, d_{secondary PVDvHC}=1.00). There were no significant differences observed between women with PVD 1 and PVD 2 on measures of depression or anxiety.

Both PVD onset subtypes reported significantly greater levels of perceived stress than HCs (d_{primary PVDvHC}=0.56, d_{secondary PVDvHC}=0.91). There was also a trend for higher levels of perceived stress in secondary PVD compared to primary PVD (d_{primary PVDvsecondary PVD}=0.37).

Compared HCs, those with secondary PVD reported higher levels of pain catastrophizing (; d_{secondary PVDvHC}=0.81); primary PVD also tend to report greater levels of catastrophizing than HC (d_{primary PVDvHC} =.46) Furthermore, secondary PVD reported significantly higher levels of pain catastrophizing than primary PVD (d_{primary PVDvsecondary PVD}=0.39).

Differences in early adverse life events and resilience

Women with secondary PVD (17.9%) and HCs (15.7%) were significantly more likely to experience sexual abuse in childhood than women with primary PVD (2.2%). A trend was also observed that women with secondary PVD (28.4%) compared to primary PVD (13.3%) were more likely to report having grown up in households with substance abusers (Table 2). Finally, women with primary PVD (46.3%) compared to HCs (20.2%) were more likely to have lived with house members who experienced mental illness; a similar trend (p<0.10) was observed for secondary PVD (32.8%). However, there were no differences in overall score on the Adverse Childhood Experiences (ACE) Scale among the PVD Subtypes and HCs.

Women with secondary PVD reported a higher score on the sexual abuse subscale than women with primary PVD (d=0.36). Also, secondary PVD experienced a higher score on the physical abuse subscale compared to HCs (d=0.45), but not primary PVD. However, no significant differences were observed among the three groups in the overall score for early adverse life events assessed with the Early Traumatic Inventory-Self Report.

Lower scores for resilience were observed for primary PVD and secondary PVD compared to healthy women (d_{primary PVDvHC}=0.59, d_{secondary PVDvHC}=0.65). No significant difference in resilience was observed between PVD onset groups.

Brain Functional Connectivity

Primary PVD compared to HCs

Compared to HCs, primary PVD showed significantly altered functional connectivity predominantly in default mode network (q<.01) and brainstem connectivity (q<.05), see Figure 2, Supp.Table 1A.

Figure 2.

Functional connectivity differences in women with primary provoked vestibulodynia compared to healthy controls. Compared to healthy controls (HCs), women with primary provoked vestibulodynia (PVD) showed significantly altered functional connectivity predominantly in default mode network and brainstem connectivity. Resting state functional connectivity analysis revealed significant differences in 91 pairwise connections, 76 pairs among cortical and subcortical brain regions (37 greater connections primary PVD> HC, 39 lower were connections, primary PVD<HC) and 15 pairs with brainstem regions (12 greater connections, 3 lower). (A) The circle graph illustrates the significant differences in resting state functional brain connectivity. Each brain region is represented as rectangle lining the circle and the colors denote brain networks. Red lines reflect significantly greater connectivity in primary PVD compared to HC while blue lines reflect significantly lower connectivity. Left and right halves of the circle correspond to cerebral hemispheres. Significance was set at q < 0.01 for cortical and subcortical regions and q < 0.05 for brainstem regions. Abbreviations: SalVentAttn=Salience/Ventral Attention, Default=Default Mode, TempParietal=Temporal/Parietal (B) Adjacency matrix depicting the total number of significantly different functional connections between each network. The darker green colors reflect higher numeric values. (C) The bar plot depicts the percentage of significant different pairwise functional connections by network from the 91 significantly different pairwise connections between women with primary PVD and HCs. The number inside the parentheses denotes the total number of unique regions of interests.

Cortical Functional Connectivity.

Intrinsic connectivity within the default mode network (DMN) was significantly (q<.01) greater in primary PVD compared to HCs predominantly in the precuneus and posterior cingulate cortex (PCC). Additional DMN regions with significantly altered intrinsic connectivity included the prefrontal cortex (PFC), inferior parietal cortex, pregenual anterior and posterior cingulate cortex (pACC, PCC), retrosplenial cortex (Rsp), parahippocampal cortex (PHC), and middle temporal cortex.

The inter-network connectivity of the DMN was altered in primary PVD compared to HCs. Compared to HCs, primary PVD had decreased connectivity between DMN regions (including the precuneus, PFC, medial temporal cortex, retrosplenial cortex, and pACC), and salience ventral attention (VTA) network regions, specifically the paracentral gyrus, medial frontal cortex, parietal operculum cortex, insula, and IPL.

Default mode and control network connectivity differed between primary PVD compared to HCs. Specifically, the dorsomedial PFC and pACC (DMN) showed lower connectivity with the lateral PFC (Control) in primary PVD. Primary PVD also showed decreased connectivity between ventromedial PFC, pACC, PCC/precuneus middle temporal cortex with supramarginal gyrus (somatosensory association cortex).

Default mode and dorsal attention network connectivity was also altered. Specifically, primary PVD had decreased connectivity between the precuneus/PCC, ventromedial PFC/ pACC, and middle temporal cortex in the DMN and the post central gyrus /supramarginal gyrus in the dorsal attention network. The precuneus (DMN) exhibited decreased connectivity with somatomotor regions including the superior temporal cortex, superior parietal cortex, and primary motor cortex in women with primary PVD.

Decreased connectivity was observed between DMN (including the precuneus, inferior parietal cortex, PFC, and superior temporal cortex) with regions in the temporal parietal network, including the lateral superior temporal cortex, supramarginal gyrus, and lingual cortex. Primary PVD also showed greater connectivity between the inferior parietal cortex and the visual network (bilateral medial occipito-temporal cortex)

Greater intrinsic connectivity was found within the SVA. Primary PVD patients demonstrated greater connectivity within the paracentral gyrus, precuneus, insula, inferior parietal cortex, and frontal operculum. Lower connectivity was found between the somatomotor (post-central and precentral gyrus) and the control network (ventromedial prefrontal cortex) in primary PVD. Subcortically, patients with primary PVD exhibited greater connectivity between the thalamus and the parahippocampal cortex compared to healthy women.

Brainstem Functional Connectivity.

Primary PVD compared to HCs showed significant (q<.05) increases in intra-brainstem connectivity including 1) locus coeruleus connectivity with the dorsal and median raphe, oral pontine reticular nucleus, and parabrachial complex and 2) the parabrachial complex with the reticularis pontis oralis (Figure 2, Suppl Table 1B). Additionally, the pedunculopontine tegmental nucleus showed greater connectivity with the superior prefrontal cortices (DMN, Control Network) and the superior parietal cortex (Dorsal Attention Network). Additionally, primary PVD showed decreased connectivity between the median raphe and the oral pontine reticular nucleus with the prefrontal cortex.

Secondary PVD compared to HCs

Relative to primary PVD, secondary PVD had fewer significant differences in functional connectivity compared to HCs. Secondary PVD subjects primarily showed differences in the dorsal attention-somatomotor network connectivity (q<.01, Figure 3, Suppl Table 2A).

Figure 3.

Functional connectivity differences in women with secondary provoked vestibulodynia compared to healthy controls. Compared to healthy controls (HCs), women with secondary provoked vestibulodynia (PVD) primarily showed differences in the dorsal attention-somatomotor network connectivity. Resting state functional connectivity analysis revealed significant differences in 29 pairwise connection, 27 connections (24 greater connections secondary PVD> HC, 3 lower connections, secondary PVD<HC) of cortical brain regions and 2 pairs with brainstem regions (2 lower connections, secondary PVD<HC) with significantly altered functional connectivity in women with secondary PVD when compared to HCs. (A) The circle graph illustrates the significant differences in resting state functional brain connectivity. Each brain region is represented as rectangle lining the circle and the colors denote brain networks. Red lines reflect significantly greater connectivity in primary PVD compared to HC while blue lines reflect significantly lower connectivity. Left and right halves of the circle correspond to cerebral hemispheres. Significance was set at q < 0.01 for cortical and subcortical regions and q < 0.05 for brainstem regions. Abbreviations: SalVentAttn=Salience/Ventral Attention, Default=Default Mode, TempParietal=Temporal/Parietal (B) Adjacency matrix depicting the total number of significantly different functional connections between each network. The darker green colors reflect higher numeric values. (C) The bar plot depicts the percentage of significant different pairwise functional connections by network from the 29 significantly different pairwise connections between women with primary PVD and HCs. The number inside the parentheses denotes the total number of unique regions of interests.

Cortical Functional Connectivity.

Compared to HCs, secondary PVD had significantly (q<.01) greater connectivity between the superior parietal cortex (dorsal attention network) and several somatomotor regions including the primary motor cortex, superior temporal gyrus, insula, and anterior transverse temporal gyrus. Greater connectivity between the superior parietal cortex and the visual network (intracalcarine and calcarine sulcus, parietal-occipital cortex, and lingual gyrus) as well as the SVA (superior frontal and inferior parietal cortices)

In the limbic network, women with secondary PVD exhibited greater connectivity in between the inferior temporal cortex (limbic network) and the precuneus (control network), as well as the insula (somatomotor network). Finally, secondary PVD patients experienced a decrease in connectivity between the precuneus (DMN) and the inferior parietal lobe/supramarginal gyrus (SVA) when compared to HCs.

Brainstem Functional Connectivity

Few significant (q<.05) differences in brainstem connectivity were observed (Figure 3, Suppl Table 2B). However, secondary PVD compared to HCs had lower connectivity between the locus coeruleus and the superior parietal cortex (dorsal attention network) as well as between the parabrachial complex (PBC) and the inferior parietal cortex (control network).

PVD Onset-dependent Differences in Functional Connectivity (secondary PVD compared to primary PVD)

Compared to primary PVD, secondary PVD subjects primarily showed differences in the dorsal attention-somatomotor network (q<.01) and brainstem connectivity (q<.05).

Cortical Functional Connectivity.

(Figure 4, Suppl Table 3A). In secondary compared to primary PVD, greater connectivity was observed primarily between the superior parietal cortex (dorsal attention network) and somatomotor regions including the primary motor cortex, primary sensory cortex, subcentral cortex, and premotor cortex (q<.01). The superior parietal cortex also showed greater intrinsic dorsal attention network connectivity with the lateral occipital/temporal cortex and posterior central gyrus/superior parietal cortex and decrease connectivity with PFC (Control) in woman with secondary PVD. Compared to primary PVD patients, secondary PVD patients also experienced greater connectivity between the superior parietal cortex within the dorsal attention network and throughout the occipital cortex (visual network). Furthermore, secondary PVD patients exhibited lower intrinsic DMN connectivity compared to primary PVD patients (specifically within the precuneus and PCC) but higher connectivity between the middle temporal cortex (DMN) and the inferior frontal cortex and dorsolateral PFC (SVA). Finally, increased connectivity was observed between the control network (dorsolateral PFC, inferior parietal cortex) and the (superior partial and primary sensory cortices).

Figure 4.

Functional connectivity differences in women with secondary compared to primary provoked vestibulodynia. Compared to primary PVD, women with secondary PVD subjects primarily showed differences in the dorsal attention-somatomotor network and brainstem connectivity. Resting state functional connectivity analysis revealed significant differences in 45 pairwise connections, 39 connections (36 greater connections secondary PVD> primary PVD, 3 lower connections, secondary PVD<primary PVD) of cortical brain regions and 6 pairs with brainstem regions (1 greater connections secondary PVD> primary PVD, 5 lower connections, secondary PVD<primary PVD) with significantly altered functional connectivity in women with secondary PVD when compared to primary PVD. (A) The circle graph illustrates the significant differences in resting state functional brain connectivity. Each brain region is represented as rectangle lining the circle and the colors denote brain networks. Red lines reflect significantly greater connectivity in primary PVD compared to HC while blue lines reflect significantly lower connectivity. Left and right halves of the circle correspond to cerebral hemispheres. Significance was set at q < 0.01 for cortical and subcortical regions and q < 0.05 for brainstem regions. Abbreviations: SalVentAttn=Salience/Ventral Attention, Default=Default Mode, TempParietal=Temporal/Parietal (A) Adjacency matrix depicting total number of significantly different functional connections between each network. The darker green colors reflect higher numeric values. (B) The bar plot depicts the percentage of significant different pairwise functional connections by network from the 45 significantly different pairwise connections between women with primary PVD and HCs. The number inside the parentheses denotes the total number of unique regions of interests.

Brainstem Functional Connectivity.

Significantly (q<.05). lower connectivity between the locus coeruleus with median raphe (MR) nucleus and the oral pontine reticular nucleus was observed in secondary PVD, compared to primary PVD (Figure 4, Suppl Table 3B). Secondary PVD also exhibited lower connectivity in the midbrain reticular formation and the periaqueductal gray (PAG) to the superior parietal cortex, a dorsal attention region. Greater connectivity was observed between the median raphe and the primary motor cortex in secondary PVD compared to primary PVD patients.

A comparison of functional connectivity between the entire PVD cohort (N=114) and HCs is provided in the supplemental materials (Suppl Results, Suppl Table 4, Suppl Figure 1).

Correlational Analysis.

Detailed results from exploratory correlational analysis between region-to-region resting state connectivity that differed between primary and secondary PVD and psychosocial and pain assessment are depicted in Suppl Figure 2 and Suppl Table 2. The main finding was that the increased dorsal attention network connectivity observed in secondary PVD compared to primary PVD and HCs showed small to moderate effect size association with a history of sexual abuse and reduced sexual functioning satisfaction scores at uncorrected p <.05. The reported associations did not remain significant after FDR correction for multiple testing.

DISCUSSION

In the present study, women with PVD showed unique RSFC and clinical profiles based on the onset time of first symptoms. Compared to HC, women with primary PVD showed significantly higher intrinsic connectivity within the default mode network and brainstem nuclei and lower connectivity between the default mode network and salience, dorsal attention, and somatomotor networks. In contrast, women with secondary PVD showed fewer differences in RSFC compared to HCs, and these were primarily enhanced connectivity between the dorsal attention and somatomotor networks and lower brainstem connectivity with cognitive control regions. Direct comparisons by onset type indicated that women with secondary PVD had increased dorsal attention-somatomotor network connectivity whereas women with primary PVD predominantly show greater intrinsic connectivity within the brainstem and the default mode network. Furthermore, women with secondary PVD compared to primary PVD reported greater incidence of early life sexual abuse, along with greater pain catastrophizing, greater 24-hour symptom unpleasantness, and less sexual satisfaction. Ultimately, maladaptive functional neuroplasticity in cortical cognitive control mechanisms involved in the central amplification pain and sensorimotor signals (dorsal attention-somatomotor connectivity) may be driving symptom presentation in secondary PVD. Furthermore, changes in the RSFC of brainstem circuitry involved in the processing and modulation of ascending and descending signals from and to the periphery is a key feature of primary PVD. The findings from the present study contribute to our understanding of supraspinal alterations driven by PVD onset subtype and may influence future treatment of primary and secondary PVD.

Greater engagement of cognitive control networks in secondary PVD

The RSFC alterations observed in secondary PVD are consistent with alterations in executive functioning networks involved in attention and cognitive control[29,39]. Women with secondary PVD compared to HCs and primary PVD exhibited greater connectivity between the superior parietal cortex of the dorsal attention network, the somatomotor and the occipital/visual networks. Furthermore, connectivity between the lateral prefrontal cortex, a cognitive control region involved in inhibition, and the superior parietal cortex region driving the differences in dorsal attention-somatomotor network connectivity was decreased. Finally, women with secondary PVD showed reduction in RSPW within the brainstem and between the brainstem and the dorsal attention network. Together, these findings suggest greater attention to spontaneous sensorimotor signals and increased endogenous pain facilitation in secondary PVD compared to primary PVD. Consistent with this interpretation, secondary PVD compared to primary PVD, reported greater pain catastrophizing, greater 24-hour symptom unpleasantness and pain intensity but not differences in evoked vulvar vestibular pain or vaginal muscle tenderness.

The observed alterations in executive network functioning may reflect a shared central mechanism across chronic pain disorders[64]. Similar increases in connectivity between the dorsal attention network and the somatomotor and occipital/visual network have been reported in irritable bowel syndrome and ulcerative colitis[94]. Functional alterations in the occipital network have also been reported in patients with chronic inflammation and with chronic pain symptoms, including migraine, chronic low back pain and fibromyalgia[53,63,82,94]. These alterations involving the occipital network are thought to reflect compromised central processing and modulation of nonpainful multisensory signals and may be a shared central mechanism across chronic pain disorders[1,64]. In the current study, there was a statistical trend for women with secondary PVD to report a greater number of somatic symptoms of discomfort or pain as well as sensory sensitivity to nonpainful environmental stimuli in the past year. These increases correlated with greater connectivity between the superior parietal cortex and the occipital network.

Women with secondary PVD compared to primary PVD but not HCs reported greater incidence of early life sexual abuse. This greater incidence of sexual abuse correlated with the observed increases in dorsal attention connectivity with the sensory and motor networks. Furthermore, increased incidence of sexual abuse was associated with the greater RSPW between the serotonergic, median raphe and the primary somatosensory cortex in secondary PVD compared to primary PVD. Alterations in median raphe nucleus (and its serotonergic projections) have been linked to vulnerability to stress-related disorders and chronic pain later in life[8,60]. Also, women experiencing childhood abuse are more likely to report vulvodynia symptoms[45,50]. Several prospective studies have linked early adverse life stressors including sexual abuse to increases in proinflammatory responses[26,49,67,93] and diagnosis of chronic pain conditions later in life[6,27,52,76,80]. Although speculative, these findings suggest a possible role for early life trauma as a vulnerability factor for the emergence of symptoms in some women with secondary PVD.

Greater connectivity in brainstem and default mode network in primary PVD.

The brainstem plays a critical role in gating pain and other sensory inputs from the body to the brain[66] and increased descending facilitation of engagement of brainstem regions have been demonstrated in preclinical models of chronic pain[65,69]. In the present study, women with primary PVD exhibited greater functional connectivity within the brainstem compared to HCS and secondary PVD. Women with primary PVD demonstrated greater functional connectivity between the following regions of interest: 1) the locus coeruleus and the median raphe nucleus, 2) the locus coeruleus and the nucleus reticularis pontis oralis. The locus coeruleus (LC) plays a crucial role in orchestrating the different components of the central stress response, including ascending noradrenergic arousal pathways, engagement of endogenous descending pain modulation pathways[19,61,92], and through its close connections with Barrington’s nucleus in the regulation of micturition. Women with primary PVD compared to secondary PVD also showed greater connectivity between the periaqueductal gray and the superior parietal cortex (dorsal attention network). The periaqueductal gray (PAG) gates pain signals from the spinal cord to the thalamus[48], but also plays a crucial role in the engagement of descending pain modulating pathways. Interestingly, the observed differences in clinical and psychosocial measures between primary PVD and secondary PVD were not associated with the observed differences in brainstem connectivity. Some studies have demonstrated greater vestibular nerve fiber density and thickness, neural hypertrophy, and vestibular progesterone receptors but less evidence of localized inflammation in primary compared to secondary PVD[35,58,72,95]. It is possible that these alterations in the vagina and vulvar vestibule may lead to increased nociceptive signaling and subsequent central sensitization of brainstem circuitry in women with primary PVD. Ultimately, the findings suggest a possible mechanistic role of the locus coeruleus complex and the periaqueductal gray in pain facilitation driving vulvar pain symptoms in primary PVD.

In line with previous findings[43], women with primary but not secondary PVD had extensive increases in connectivity within the default mode network when compared with HCs with fewer differences observed when compared with secondary PVD. The default mode network is involved in passive self-referential mental activity and prospection, and in the evaluation and recollection of previous experiences[20,75]. Alterations in the default mode network connectivity have been consistently observed across chronic pain conditions[4,7,62]. The default mode network and the dorsal attention network have been shown to be anticorrelated, with increased activity in the internally directed default mode network leading being associated with higher externally directed attentional network activity and vice versa[6,33,44]. Ultimately, the findings support the hypothesis that women with secondary PVD show greater evidence for central amplification of sensory signals whereas women with primary PVD show evidence for central sensitization in brainstem circuitry responsible for the processing and modulation of ascending and descending peripheral signals.

Implications for therapy

Findings from the present study provide evidence of differing central mechanisms driving primary and secondary PVD with the potential to inform development of more effective therapeutic approaches for women suffering with PVD. The brain alterations suggest that women with primary PVD may respond more to centrally targeted therapies such as cognitive behavioral therapy, hypnosis and mindfulness as well as centrally targeted medications, whereas secondary PVD may respond more to peripherally targeted therapies aimed at reducing increased peripheral nociceptive signaling[10,14,22,24,38,47].

Strengths and Limitations

The atlas-based investigation of individual brainstem nuclei rather than relying on one region of interest for the entire brainstem represents an important contribution and revealed brainstem nuclei specific functional connectivity differences. However, precise neuroanatomical localization of these the brainstem nuclei is limited due to the limited spatial resolution of the 3T scanner. Furthermore, the current sample size did not have adequate power to detect effect size differences less than Cohen’s d =.50. Finally, a more formal diagnostic assessment for the presence of concurrent comorbid chronic pain conditions is important to control for the influence of comorbid pain conditions more accurately. Finally, the cross-section study design only permits correlational not causal inferences. Ultimately, a longitudinal interventional study incorporating both brain and multi-omics assessments including cytokines, mi-RNA expression, and metabolites (hormone and sphingolipid signaling pathways)[23,36,54,55] is essential to elucidate the interaction between peripheral and central mechanisms contributing to the onset and maintenance of PVD symptoms.

Conclusion

We have shown that women with secondary PVD and primary PVD differ from HCs and from each other. Women with secondary PVD show enhanced dorsal attention-somatomotor network connectivity whereas women with primary PVD predominantly show greater intrinsic RSFC within the brainstem and the default mode network. Together with the observed clinical differences, these findings have possible implications for a better understanding of subtypes in other chronic pain conditions, and for the optimization of treatments.