Fibromyalgia (FM) is a chronic pain condition that predominantly affects women. Evidence implies that FM is associated with dysfunction of the central nervous system (CNS). In this study, we investigated the structural differences between FM patients and healthy control (HC) subjects using a multivariate approach. Source-based morphometry (SBM) was applied to structural magnetic resonance imaging (sMRI) data consisting of 20 FM patients (46.4 ± 12.5), and age and gender-matched 20 HC subjects (42.1 ± 12.5). SBM revealed greater grey matter volume (GMV) in the bilateral thalamus in FM patients. Conversely, lower GMV was found in the bilateral putamen, bilateral pallidum, right cerebellum, right calcarine, right amygdala, and bilateral insula in FM patients. Further analysis indicated that grey matter deficits in the pallidum were significantly associated with pain catastrophizing, pain magnification, rumination, and feelings of helplessness, suggesting a link between structural brain changes and clinical pain metrics. These findings provide new insights into the neurobiological underpinnings of FM, highlighting the role of specific brain regions in pain processing and emotional regulation. The results underscore the potential for targeted therapeutic interventions that address both the neurobiological and psychological aspects of FM, paving the way for more effective management strategies for this complex condition.
Fibromyalgia is a chronic pain condition with a complex etiology that predominantly affects women and is marked by extensive musculoskeletal pain along with sleep, memory, fatigue, and mood issues, affecting between 2 and 3% of people worldwide1,2, yet the underlying mechanisms contributing to its pathology remain inadequately understood3. This lack of clarity continues to pose significant challenges for effective diagnosis and treatment, leading to a pressing need for advanced neuroimaging techniques that can elucidate the neural correlates of FM.
Conventional neuroimaging approaches have highlighted functional abnormalities in the processing of pain in FM patients, showcasing hyperactivity in response to nociceptive stimuli4,5. However, there is an essential gap in our understanding of the structural brain changes associated with FM. While voxel-based morphometry (VBM)6 has been employed previously to study deficits of grey matter volume in several disorders either on a voxel-by-voxel or seed-to-voxel scale7,8, the advanced multivariate technique of Source-Based Morphometry (SBM) presents a more refined methodology for detecting grey matter volumetric alterations9. SBM integrates data from multiple voxels to identify covarying “networks” and conducts statistical analysis on the covariation of these networks across subjects, rather than testing each voxel individually9. SBM uses independent component analysis (ICA)10 to decompose imaging data into maximally independent components, thereby providing a detailed landscape of structural abnormalities across several brain regions and identifying patterns of common variation between two groups9,11. This capability is crucial for identifying specific areas impacted by FM, beyond what traditional methods can reveal.
Moreover, cognitive factors such as pain catastrophizing—characterized by an exaggerated emotional response to perceived pain—have been shown to further complicate the clinical presentation of fibromyalgia12. Research indicates that pain catastrophizing not only intensifies pain perception but is also linked to structural brain changes, particularly within regions implicated in emotional regulation and pain processing13,14. Herein, we aim to investigate the relationship between grey matter volume (GMV) alterations associated with FM and the cognitive-emotional dimension of pain catastrophizing.
This study specifically targets the structural differences in the brains of FM patients compared to healthy controls, using SBM as a robust technique for identification of these differences. By examining how variations in GMV relate to pain catastrophizing and its subscales—such as rumination, magnification, and helplessness—we intend to elucidate the neurobiological underpinnings of FM. Understanding these relationships may provide new avenues for targeted therapeutic interventions, balancing both pharmacological and psychological treatment strategies tailored to the individual needs of FM patients. This represents not only a significant advancement in our understanding of the disorder but also an innovative step towards enhanced management strategies that could substantially improve patient quality of life. We hypothesized that SBM would reveal robust and significant alterations in GMV in FM patients.
Materials and methods
Participants
Data from 20 FM patients and 20 healthy controls (HC) subjects were obtained from a public data set (Available at: https://openneuro.org/datasets/ds001928/versions/1.1.0)15. FM patients were recruited from the Hospital General of the Secretaria de Salud and an FM help group, both in the city of Queretaro, Mexico. The inclusion criteria for FM individuals were: (1) age 18 years or older, (2) female, (3) meeting the American College of Rheumatology (ACR) 1990 classification criteria and the ACR 2010 diagnostic criteria for FM by a trained rheumatologist, (4) right-handed, and (5) spontaneous and continuous pain in daily life (Visual Rating Scale [VRS] > 5, average of a month). The exclusion criteria were: (1) inability to move or walk, (2) uncontrolled endocrine problems, (3) neurological diseases (e.g., stroke, epilepsy, traumatic brain injury), (4) auditory problems, (5) pregnancy and/or breastfeeding, and (6) MRI contraindications. HC subjects were enrolled and age-matched to the recruited FM participants. The inclusion criteria were: (1) healthy adult females, (2) right-handed. The exclusion criteria were: (1) acute or chronic pain (e.g., osteoarthritis), (2) pregnancy and/or lactating women, and (3) MRI contraindications. Written informed consent was obtained from each patient before the study, which was conducted following the Declaration of Helsinki. Patients received no compensation for taking part in the study. Ethical permission was obtained from the Bioethics Committee of the Institute of Neurobiology, UNAM Juriquilla, Queretaro, Mexico . FM patients were asked not to use painkillers on the day of testing. HC subjects were screened to ensure none of them experienced any type of pain on the day of testing.
To evaluate pain before MRI scanning, pain intensity was measured only in FM patients, using a verbal rating scale (0 = no pain, 10 = worst pain possible). Both FM and HC subjects answered the following questionnaires: the Spanish version of the Pain Catastrophizing Scale (PCS) to measure thoughts and feelings when experiencing pain16, State-Trait Anxiety Inventory (STAI) to measure both immediate (state) and broad (trait) emotional, cognitive, and behavioral elements of anxiety17, and Center for Epidemiologic Studies Depression Scale (CESD), a questionnaire for depression and depressive disorder18. Helplessness, pain magnification, rumination, pain self-perception, and FM years were also recorded for symptom assessment.
MRI acquisition
The image acquisition was performed with a 3-T GE Discovery MR750 scanner (HD, GE Healthcare, Waukesha, WI, USA) and a commercial 32-channel head coil array. High-resolution T1-weighted anatomical images were obtained using the FSPGR BRAVO pulse sequence with the following parameters: plane orientation = sagittal; TR = 7.7 ms; TE = 3.2 ms; flip angle = 120; matrix = 256 × 256; FOV = 256 mm2; slice thickness = 1.1 mm; number of slices = 168; gap = 0 mm; slice order = interleaved.
Image preprocessing
High-spatial-resolution T1-weighted MR imaging data were processed using the Computational Anatomy Toolbox (CAT12) in statistical parametric mapping software (SPM12). All T1 images were first checked for artifacts and reoriented to adjust image origins at the anterior commissure. Segmentation was then done to separate the T1 images into grey matter (GM), white matter (WM), and cerebrospinal fluid (CSF) and resampled to a volume image resolution of 3 × 3 × 3 mm3. After the data quality and sample homogeneity check, the segmented GM images were smoothed using an 8-mm full width at half maximum Gaussian kernel. Total intracranial volume (TIV) was estimated for each subject. The smoothed GM images were used for subsequent analyses.
Source based-morphometry (SBM) analysis
First, we used the Minimum Description Length (MDL) criterion to estimate the number of components (K)19, which resulted in K = 4 components.
All participants’ pre-processed grey matter images were processed using SBM as implemented in the GIFT toolbox SBM was performed using the Infomax algorithm20. This process was repeated 20 times within the ICASSO algorithm to enhance component reliability and consistency21. All participants’ data were converted into one 40-row participant-by-grey matter data matrix “measure matrix” (40 participants × 698,228 voxels) in which each row is a vectorized image of grey matter from each participant, with the voxels from the 3D image unwrapped into a single row. This matrix was then decomposed into a participant-by-component “mixing matrix” and a component-by-voxel “source matrix”. The mixing matrix expresses the relationship between the 40 subjects and the 4 components, with its rows indicating the degree to which the components contribute to each subject while the columns indicate how each component contributes to the 40 subjects (40 subjects × 4 components). The source matrix however expresses the relationship between the voxels in the brain and the 4 components, with its rows indicating how each component contributes to different brain voxels while the columns indicate the contribution of one brain voxel to each component (4 components × 698,228 voxels). Finally, a two-sample t-test was performed on loading coefficients in the mixing matrix to identify components with significant differences between the FM and HC groups.
Correlation analysis
To determine the relationship between abnormal grey matter regions and FM symptoms metrics, we extracted regions that showed significant grey matter differences between FM patients and HC subjects (voxels from the significant ROIs were extracted from all subjects) and performed a partial correlation analysis using age as a covariate. FM metrics used include; FM years, rumination, pain magnification, STAI scores, CESD scores, pain intensity, pain catastrophizing scale scores, pain self-perception scores, and helplessness. GMV values were residualized and used for the correlation analysis in order to accurately reflect the linear relationship between the variables.
Statistical analysis
Statistical analysis was done using the mixing matrix. Since the columns of the mixing matrix indicate how each component contributes to the 40 subjects, we applied a two-sample t-test to every column of the mixing matrix to determine which components show significant differences between FM patients and HC subjects. Age and TIV were regressed out as covariates to control for their potential confounding effects. False discovery rate (FDR) was used for multiple comparison correction, thresholding at p < 0.05. The identified spatial component maps, representing areas with significant differences between FM patients and HC subjects, were visualized with a threshold of |Z
Results
Demographic and clinical characteristics
Our study comprised 20 FM patients and 20 HC subjects matched for age and sex. The demographic and clinical information of the participants are shown in Table 1. The two groups did not differ significantly in age. FM patients had a mean disease duration of 5.2 ± 5.0 years and a mean pain intensity of 7.2 ± 1.6 indicating they were in severe pain. Compared to healthy controls, FM patients demonstrated higher scores in pain catastrophizing, helplessness, pain magnification, rumination, pain self-perception, anxiety, and depressive symptoms.
Source based-morphometry
Based on the MDL criterion, the grey matter images of all participants were decomposed into 4 components (Supplementary Fig. 1). All 4 components were stable with no artifacts. Among the 4 components, only component 4 showed significant differences in loading coefficients between FM patients and HC subjects. The patient group showed higher loading coefficients in this component than the HC group (component 4: t = 6.9932, FDR corrected P = 0.0002). Components 1, 2, and 3 did not show significant loading coefficients between the 2 groups (component 1: t = − 1.22, P = 0.22627; component 2: t = 0.71446, P = 0.47931; component 3: t = − 0.53863, P = 0.59328). However, the interpretation of loading coefficients depends on the sign of the corresponding spatial map. For instance, if the FM group has higher loading coefficients than the HC group for a component with a predominantly positive spatial map, the FM group exhibits greater tissue volume in this pattern. Conversely, a predominantly negative spatial map indicates rather lower tissue volume in the FM group. The significant component (component 4) representing distinct grey matter regions between the two groups was threshold at |Z|> 3 (Fig. 1). Specifically, the bilateral thalamus showed greater GMV in the FM group while lower GMV in the right cerebellum lobule 6, bilateral putamen, bilateral pallidum, bilateral insula, right amygdala, right hippocampus, and the right calcarine were observed in the FM group. Detailed information about these brain regions is presented in Supplementary Table 1. The insignificant components maps were also threshold at |Z|> 3. Component 1 mainly showed greater GMV in the cerebellum of FM patients compared to HC subjects (Supplementary Fig. 2 and Supplementary Table 2). Component 2 exhibited greater GMV in the bilateral thalamus in FM compared to HC subjects (Supplementary Fig. 3 and Supplementary Table 3). In component 3, the bilateral temporal middle pole, the right hippocampus, left putamen, right fusiform, and left inferior temporal gyrus showed greater GMV in FM patients whiles the right cerebellum lobules 6 and 8 showed lower GMV in FM patients (Supplementary Fig. 4 and Supplementary Table 4).
