7T multimodal MRI reveals structural-functional-quantitative susceptibility mapping abnormalities of new daily persistent headache Authors: Xiu Liu, Xinyu Wang, Hui Su, Mingjie Zhang, Xiangbing Bian, Caohui Duan, Song Wang, Chenghui Pi, Xiting Nie, Wei Dai, Yin Sun, Longteng Ma, Shuqing Wang, Siyuan Xie, Deqi Zhai, Yuhan Wu, Xin Lou, Zhao Dong Published: 2025 ABSTRACT BackgroundNew daily persistent headache (NDPH) is a rare, refractory primary headache with an unclear pathophysiological mechanism. Previous neuroimaging studies on NDPH have been largely limited to 3T MRI, which fails to thoroughly reveal microstructural changes, particularly subregional abnormalities and iron metabolism alterations. Based on this, the present study employs 7T multimodal imaging techniques to investigate structural, functional, and iron metabolism abnormalities in the whole brain and, in particular, the changes in subregions of the limbic system.MethodsA total of 23 individuals with NDPH and 23 healthy controls (HCs) underwent 7T MRI, including T1-weighted three-dimensional magnetization-prepared 2 rapid acquisition gradient echo (3D-T1WI-MP2RAGE) and resting-state fMRI (rs-fMRI); among these patients, 19 also underwent quantitative susceptibility mapping (QSM). Structural volumes, functional metrics (fractional amplitude of low-frequency fluctuations [fALFF], regional homogeneity [ReHo]), and iron deposition (assessed via QSM) were analyzed, and their correlations with clinical parameters (e.g., headache history, anxiety/depression scores) were examined.ResultsCompared to HCs, individuals with NDPH exhibited significantly less volume in the right accumbens area and left caudal anterior cingulate cortex after false discovery rate (FDR) correction. Widespread changed fALFF and ReHo values were observed, with correlations to clinical features. QSM values were decreased in right paracentral, right cuneus and left precentral with increasing in left rostral middle frontal.Conclusion7T multimodal MRI identifies widespread structural, functional, and iron metabolism abnormalities in NDPH, particularly in limbic subregions, highlighting a “pain-emotion” interaction mechanism. These findings provide preliminary insights into NDPH pathogenesis.Supplementary InformationThe online version contains supplementary material available at 10.1186/s10194-025-02210-0. Introduction New Daily Persistent Headache (NDPH) is an extremely rare primary disabling headache. According to the International Classification of Headache Disorders, 3rd edition (ICHD-3) established by the International Headache Society (IHS), NDPH is defined as “ distinct and clearly remembered onset, with pain becoming continuous and unremitting within 24 h; present for >3 months; not better accounted for by another ICHD-3 diagnosis” [1]. Individuals diagnosed with NDPH can precisely describe the moment when their headache first appeared [2]. Previous studies have shown that the prevalence of new daily headache in the population is 0.01%-0.3% [3, 4], and it is recognized as one of the chronic daily headache(CDH) [5]. Moreover, NDPH is increasingly recognized as a heterogeneous syndrome rather than a single disorder, with recent cluster analyses identifying at least three distinct phenotypes [6]. 7 tesla magnetic resonance imaging (7T MRI), leveraging its superior magnetic field homogeneity and signal sensitivity, breaks through the limitations of traditional low-field devices and pioneers a new dimension for neuroimaging research. Compared with the clinically commonly used 3T equipment, 7T MRI achieves a remarkable improvement in spatial resolution, enabling the capture of more subtle structural alterations [7]. In previous studies by our team, 7T high-field MRI was utilized to analyze the unique imaging feature changes in chronic migraine patients with medication overuse (CM + MOH) compared to those without medication overuse (CM). For instance, CM + MOH patients exhibited a distinct pattern of less gray matter volume in brain regions associated with memory and visual processing, accompanied by obvious white matter disruption [8]. Although a substantial number of studies have explored the pathogenesis of NDPH using multimodal MRI and yielded significant findings, all these investigations were based on 3T MRI [9, 10]. However, there are obvious inconsistencies in their conclusions. For instance, Szabo et al. observed brain structural changes in adolescent individuals with NDPH [9], whereas Naegel et al. failed to detect significant brain structural abnormalities [10]. Given the complex anatomical structure of the nervous system, it is necessary to further explore the microstructural changes in the brain, particularly at the subregional level. Additionally, current research has not analyzed the changes in NDPH on quantitative susceptibility mapping (QSM) sequences. As an advanced MRI technique, QSM quantifies iron deposition in brain tissue through a non-invasive approach, playing a prominent role in the research of various neurological diseases, such as Alzheimer’s disease [11], multiple sclerosis [12] and Parkinson’s disease [13]. Previous studies have shown abnormal iron deposition patterns in chronic migraine and medication overuse headache [14, 15]. However, the iron deposition pattern in NDPH has not been reported. The limbic system serves as a core neural network integrating emotion, memory, and somatic sensation [16, 17]. Its predominant constituent structures—such as the thalamus (relaying pain signals and processing emotions), hypothalamus (regulating stress), hippocampus (encoding pain memory), and amygdala (processing emotions)—modulate pain and emotion through complex neural circuits [18–22]. As a chronic primary headache, NDPH is characterized not only by intractable pain but also by frequent comorbidities of emotional disorders such as anxiety and depression [23]. The vicious cycle between pain and emotion significantly impairs patients’ quality of life [21, 24]. Given the critical role of the limbic system in pain-emotion regulation and the abnormal affect-pain interaction observed in individuals with NDPH, in-depth investigation of limbic-related brain regions in the pathological mechanism of NDPH will provide essential theoretical foundations for deciphering its pathogenesis and developing precision therapeutic strategies. This study enrolled individuals with NDPH and healthy controls (HCs) for 7T MRI analysis. It acquired 7T T1-weighted three-dimensional magnetization-prepared 2 rapid acquisition gradient echo (3D-T1WI-MP2RAGE) and resting-state fMRI (rs-fMRI) data and QSM sequences which aimed to clarify the structural, functional changes and abnormal iron deposition patterns in brain regions, especially the limbic system (hypothalamus, thalamus, amygdala, hippocampus), so as to provide references for the exploration of NDPH pathogenesis and attempt to propose new therapeutic targets. Materials and methods Participants This study was approved by the Ethics Committee of Chinese PLA General Hospital (S2021-700-01, Beijing, China). All research procedures adhered to the latest revised Helsinki Declaration. Participants were recruited from Chinese PLA General Hospital and provided written informed consent after a comprehensive explanation of the study protocol. In this cross-sectional study, we enrolled 23 individuals with NDPH and 23 HCs to investigate differences in structure, function, and iron deposition patterns. Study participants were either newly diagnosed cases or patients who had discontinued previously ineffective treatments at least 2 weeks as reported before [9]. They did not take prophylactic medications before scanning (after completing MRI examinations, personalized treatments were administered based on patients’ conditions). Each participant completed standardized neuropsychological assessments and underwent 7T 3D-T1, rs-fMRI, and some of them underwent QSM scans. Additionally, two neurologists obtained diagnostic information and detailed medication histories through structured interviews. All participants were right-handed. Inclusion criteria for individuals with NDPH were based on the ICHD-3 [1], including: (1) age 18–60 years; (2) diagnosis of NDPH meeting ICHD-3 criteria (A–D); (3) no comorbid neurological or psychiatric disorders (including other primary headache or pain disorders); (4) no obvious structural lesions detected by routine MRI. Exclusion criteria were: (1) other neurological diseases (e.g., epilepsy, stroke, traumatic brain injury); (2) meeting diagnostic criteria for other primary or secondary headaches; (3) history of caffeine, nicotine, or alcohol abuse; (4) contraindications to MRI. In addition to basic demographic data, clinical variables were recorded, such as Patient Health Questionnaire-9(PHQ-9), Generalized Anxiety Disorder-7(GAD-7) and Pittsburgh sleep quality index (PSQI). Inclusion criteria for HCs were: age 18–60 years, no history of primary headache, chronic pain, epilepsy, or other neurological diseases, and no history of psychiatric disorders such as depression or anxiety; no contraindications to MRI (e.g., metal implants, claustrophobia), and confirmed normal physical and psychological status through medical screening. Image acquisition All MRI scans were conducted on a whole-body 7T scanner (MAGNETOM Terra, Siemens Healthcare, Erlangen, Germany) outfitted with an 8-channel transmit coil and a 32-channel receive coil. Participants were positioned in a supine posture within the scanner bore, with foam padding applied to reduce motion-induced artifacts. For the rs-fMRI scan, echo-planar imaging (EPI) was employed, with parameters as follows: voxel dimensions = 1.8 × 1.8 × 1.8 mm³, repetition time (TR) = 2000 ms, echo time (TE) = 24 ms, flip angle = 90°, field of view (FOV) = 216 × 216 mm, matrix size = 120 × 120, slice thickness = 1.8 mm, and 80 interleaved slices. The MRI protocol also included a high-resolution 3D-T1 weighted structural sequence, with two sets of parameters used: voxel size = 0.7 × 0.7 × 0.8 mm³ (or 0.75 × 0.75 × 0.75 mm³), TR = 6000 ms (or 4300 ms), TE = 2.21 ms (or 2.27 ms), flip angle = 4° (or 5°), FOV = 224 × 224 mm, slice thickness = 0.75 mm, sagittal slices = 192 (or 240). QSM scans were performed with the following settings: voxel size = 0.6 × 0.6 × 0.8 mm³, TR = 27 ms, TE = 5.6 ms, FOV = 218 × 218 mm, slice thickness = 0.8 mm, and 160 sagittal slices. Throughout the scanning session, participants were instructed to stay relaxed with their eyes closed, and post-scan assessments confirmed that all remained awake to ensure data integrity. Data preprocessing FreeSurfer (version 7.4.1) was used for preprocessing 3D-T1 data and volumetric calculations of subregional brain structures (e.g., bilateral thalami, amygdalae, hippocampi) [25]. The standard FreeSurfer pipeline includes skull stripping, spatial normalization, segmentation of cortical and subcortical gray matter, estimation of cortical thickness, and other procedures [26–38]. Briefly, the processing steps involved first preparing the data: raw DICOM images were converted to NIfTI format, and 3D-T1 data underwent denoising using DenoiseImage (given the higher noise levels in 7T imaging). Next, the FreeSurfer “recon-all” command was executed to automate processing of each participant’s MRI data, including skull stripping, Talairach transformation, segmentation of cortical and subcortical regions, measurement of cortical thickness, and surface reconstruction. Quality control was then performed by visually inspecting segmentation results using Freeview to verify accuracy, followed by extraction of volumes for target structures. Since the standard FreeSurfer atlas lacks hypothalamic subregions, a previously published hypothalamic subregion atlas was utilized, and H-SynEx methods were applied to extract volumes of hypothalamic subregions [39]. Rs-fMRI data were preprocessed using the DPABI toolbox [40] in MATLAB R2022b, with standardized procedures adopted to ensure data quality and reproducibility. First, DICOM images were converted to NIfTI format, and the first five time points were removed to mitigate magnetic saturation effects. Preprocessing of the remaining 235 time points included slice timing correction, head motion correction (reoriented to the anterior commissure–posterior commissure plane), realignment, and coregistration with the T1-weighted structural image (which was segmented using DARTEL [41]). Subsequent steps involved spatial normalization to the Montreal Neurological Institute (MNI) template and spatial smoothing—a 6-mm full width at half maximum (FWHM) Gaussian kernel was applied for amplitude of low-frequency fluctuations (ALFF) analysis, while regional homogeneity (ReHo) analysis was conducted without smoothing to preserve local features. ALFF was calculated via fast Fourier transform (FFT) within the 0.01–0.08 Hz range and converted to z-scores to generate mean ALFF (mALFF) data; ReHo was computed using Kendall’s W to measure the local consistency of each voxel with its 27 neighboring voxels, followed by conversion to z-scores. Nuisance covariates, including white matter signal, cerebrospinal fluid signal, and head motion parameters (via Friston’s 24-parameter model), were regressed out, followed by linear detrending and bandpass filtering (0.01–0.08 Hz). All preprocessed data underwent visual inspection by two experienced radiologists to confirm quality, and the resulting fractional ALFF (fALFF) and ReHo maps were saved in NIfTI format. After exporting images from the MRI scanner in DICOM format, QSM post-processing was carried out using STI Suite software (Duke University). The process began with converting DICOM files to NIfTI format, followed by merging magnitude images from different TEs into a 4D magnitude dataset and phase images from different TEs into a 4D phase dataset. Phase images were then converted to the [-π, π] range, and QSM reconstruction was executed with the STIsuite. We employed ANTs for pairwise registration, transforming individual T1-weighted spaces generated by FreeSurfer to the MNI152 space. This method ensured precise spatial alignment between structural definitions and functional metrics while accounting for individual neuroanatomical variations. Subregions that were excessively small and lost during mask registration were excluded from subsequent region-of-interest (ROI) analyses. Statistical analysis We assessed the normality of continuous variables using the Shapiro–Wilk test. The clinical characteristics of patients with NDPH were compared with those of HCs as follows: independent sample t-tests were used for normally distributed continuous variables, Mann–Whitney U tests for non-normally distributed continuous variables, and chi-square tests for categorical variables. We used the Generalized Linear Model (GLM) to identify brain region differences with GAD-7, PHQ-9 and PSQI included as covariates for correction, and the False Discovery Rate (FDR) method was applied for multiple comparison correction. Additionally, partial correlation analyses were conducted between the differential brain regions and clinical parameters. To account for potential confounding variables, covariates such as age, sex, PSQI scores, GAD-7, and PHQ-9 were included. Statistical significance was set at P < 0.05, and all analyses were conducted in R software (version 4.3.0). Results Demographic information and clinical characteristics This study enrolled 23 individuals with NDPH and 23 HCs. There was no significant difference in age and sex between two groups. Both patients and HCs completed the assessments of GAD-7, PSQI, and PHQ-9 scales. The specific demographic information and clinical characteristics are shown in Table 1. However, since QSM sequences were only acquired from 19 of the individuals with NDPH, we reanalyzed the demographic information and clinical characteristics, as detailed in Table 2. Table 1Demographic information and clinical characteristics in structural and functional analysisHC(n = 23)NDPH(n = 23) P Age (median [IQR])30.00[26.00, 47.00]30.00[20.00, 47.00]0.2911Male (n)15180.5125Headache history (months)NA24.00 [12.00, 78.00]VAS (0–10)NA5.48 (1.47)Gad-7 (median [IQR])4.00 [0.00, 7.00]8.00 [4.00, 12.50]0.0070PHQ-9 (median [IQR])4.00 [0.00, 9.00]11.00 [8.00, 15.50]0.0013PSQI (mean (SD))6.35 (4.34)10.17 (4.79)0.0069Abbreviations: VAS, Visual Analogue Scale; PHQ-9, Patient Health Questionnaire-9; GAD-7, Generalized Anxiety Disorder; PSQI, Pittsburgh sleep quality index Table 1Demographic information and clinical characteristics in structural and functional analysisHC(n = 23)NDPH(n = 23) P Age (median [IQR])30.00[26.00, 47.00]30.00[20.00, 47.00]0.2911Male (n)15180.5125Headache history (months)NA24.00 [12.00, 78.00]VAS (0–10)NA5.48 (1.47)Gad-7 (median [IQR])4.00 [0.00, 7.00]8.00 [4.00, 12.50]0.0070PHQ-9 (median [IQR])4.00 [0.00, 9.00]11.00 [8.00, 15.50]0.0013PSQI (mean (SD))6.35 (4.34)10.17 (4.79)0.0069Abbreviations: VAS, Visual Analogue Scale; PHQ-9, Patient Health Questionnaire-9; GAD-7, Generalized Anxiety Disorder; PSQI, Pittsburgh sleep quality index Demographic information and clinical characteristics in structural and functional analysis Abbreviations: VAS, Visual Analogue Scale; PHQ-9, Patient Health Questionnaire-9; GAD-7, Generalized Anxiety Disorder; PSQI, Pittsburgh sleep quality index Table 2Demographic information and clinical characteristics in QSM analysisHC(n = 23)NDPH(n = 19) P Age (median [IQR])30.00[26.00, 47.00]30.00 [20.00, 49.00]0.4327Male (n)15140.7984Headache history (months)NA24.00 [12.00, 78.00]VAS (0–10)NA5.32 (1.49)Gad-7 (median [IQR])4.00 [0.00, 7.00]8.00 [4.00, 11.00]0.0112PHQ-9 (median [IQR])4.00 [0.00, 9.00]11.00 [8.00, 15.00]0.0028PSQI (mean (SD))6.35 (4.34)10.37 (4.56)0.0059Abbreviations: VAS, Visual Analogue Scale; PHQ-9, Patient Health Questionnaire-9; GAD-7, Generalized Anxiety Disorder; PSQI, Pittsburgh sleep quality index. QSM, quantitative susceptibility mapping Table 2Demographic information and clinical characteristics in QSM analysisHC(n = 23)NDPH(n = 19) P Age (median [IQR])30.00[26.00, 47.00]30.00 [20.00, 49.00]0.4327Male (n)15140.7984Headache history (months)NA24.00 [12.00, 78.00]VAS (0–10)NA5.32 (1.49)Gad-7 (median [IQR])4.00 [0.00, 7.00]8.00 [4.00, 11.00]0.0112PHQ-9 (median [IQR])4.00 [0.00, 9.00]11.00 [8.00, 15.00]0.0028PSQI (mean (SD))6.35 (4.34)10.37 (4.56)0.0059Abbreviations: VAS, Visual Analogue Scale; PHQ-9, Patient Health Questionnaire-9; GAD-7, Generalized Anxiety Disorder; PSQI, Pittsburgh sleep quality index. QSM, quantitative susceptibility mapping Demographic information and clinical characteristics in QSM analysis Abbreviations: VAS, Visual Analogue Scale; PHQ-9, Patient Health Questionnaire-9; GAD-7, Generalized Anxiety Disorder; PSQI, Pittsburgh sleep quality index. QSM, quantitative susceptibility mapping Changes in the volumes of cerebral cortex and subcortical nuclei across the entire brain, and analysis of their clinical correlations Compared with HCs, individuals with NDPH showed significantly less volume in the right accumbens area and left caudal anterior cingulate cortex (cACC) after FDR correction. No correlation was found between the above-mentioned brain regions and clinical indicators (detailed in Table S1). Group differences in functional values and association with clinical variables Results showed that compared with HCs, individuals with NDPH exhibited widespread and significant alterations in brain function, with numerous brain regions showing increased fALFF values after FDR correction. These brain clusters are predominantly located in the frontal lobe, as well as the parietal lobe, temporal lobe, occipital lobe, and cerebellum. Additionally, they involve the limbic system, such as the bilateral amygdala and hippocampus (detailed in Supplementary Table 2). Correlation analysis with clinical indices revealed that fALFF values in right accumbens area, right hippocampus and left lateral occipital were positively correlated with GAD-7 scores. Additionally, fALFF values in left transversetemporal was significantly negatively correlated with VAS scores. In the correlation analysis with PHQ9 scores, we found that the fALFF values of the left frontal pole was significantly positively correlated with PHQ9 scores. Moreover, we found that the fALFF values in certain brain regions, such as the left inferior parietal cortex were negatively correlated with headache history (detailed in Supplementary Table 5 and Fig. 1). All correlation analyses showed P < 0.05. Fig. 1Correlation between brain regions in fALFF and clinical scales. A. A significant positive correlation between fALFF values of the Right accumbens area and GAD-7 scores. B. A significant positive correlation between fALFF values of the Right hippocampus and GAD-7 scores. C. A significant positive correlation between fALFF values of the Left lateral occipital and GAD-7 scores. D. A significant negative correlation between fALFF values of the Left parahipoocampal and headache history (months). E. A significant negative correlation between fALFF values of the Left superior parietal and headache history (months). F. A significant negative correlation between fALFF values of the Left inferior parietal and headache history (months). G. A significant negative correlation between fALFF values of the Left supramarginal and headache history (months). H. A significant negative correlation between fALFF values of the Right transversetemporal and headache history (months). I. A significant negative correlation between fALFF values of the Right posteriorcingulate and headache history (months). J. A significant negative correlation between fALFF values of the Left parsopercularies and headache history (months). K. A significant positive correlation between fALFF values of the Left frontal pole and PHQ-9 scores. L. A significant negative correlation between fALFF values of the Left transversetemporal and VAS scores Correlation between brain regions in fALFF and clinical scales. A. A significant positive correlation between fALFF values of the Right accumbens area and GAD-7 scores. B. A significant positive correlation between fALFF values of the Right hippocampus and GAD-7 scores. C. A significant positive correlation between fALFF values of the Left lateral occipital and GAD-7 scores. D. A significant negative correlation between fALFF values of the Left parahipoocampal and headache history (months). E. A significant negative correlation between fALFF values of the Left superior parietal and headache history (months). F. A significant negative correlation between fALFF values of the Left inferior parietal and headache history (months). G. A significant negative correlation between fALFF values of the Left supramarginal and headache history (months). H. A significant negative correlation between fALFF values of the Right transversetemporal and headache history (months). I. A significant negative correlation between fALFF values of the Right posteriorcingulate and headache history (months). J. A significant negative correlation between fALFF values of the Left parsopercularies and headache history (months). K. A significant positive correlation between fALFF values of the Left frontal pole and PHQ-9 scores. L. A significant negative correlation between fALFF values of the Left transversetemporal and VAS scores During the ReHo analysis, we observed changes across a wide range of brain regions after FDR correction. Specifically, the ReHo values were significantly increased in left supramarginal, left pars triangularis, left inferior parietal and so on. In contrast, the ReHo values were significantly decreased in the right pallidum, left ventral DC, right parahipoocampal and so on. The rest data can be seen in Table S3. In the correlation analysis, the ReHo values of the left parahippocampal, left superior parietal, left inferior parietal, left supramarginal, and left parsopercularies were negatively correlated with headache history. The ReHo value of the left frontal pole was positively correlated with PHQ-9 scores (detailed in Table S6 and Fig. 2). Fig. 2Correlation between brain regions in ReHo and clinical scales. A. A significant negative correlation between ReHo values of the Left parahipoocampal and headache history (months). B. A significant negative correlation between ReHo values of the Left superior parietal and headache history (months). C. A significant negative correlation between ReHo values of the Left inferior parietal and headache history (months). D. A significant negative correlation between ReHo values of the Left supramarginal and headache history (months). E. A significant negative correlation between ReHo values of the Left parsopercularies and headache history (months). F. A significant positive correlation between ReHo values of the Left frontal pole and PHQ-9 scores Correlation between brain regions in ReHo and clinical scales. A. A significant negative correlation between ReHo values of the Left parahipoocampal and headache history (months). B. A significant negative correlation between ReHo values of the Left superior parietal and headache history (months). C. A significant negative correlation between ReHo values of the Left inferior parietal and headache history (months). D. A significant negative correlation between ReHo values of the Left supramarginal and headache history (months). E. A significant negative correlation between ReHo values of the Left parsopercularies and headache history (months). F. A significant positive correlation between ReHo values of the Left frontal pole and PHQ-9 scores Group differences in quantitative susceptibility mapping During the QSM analysis, 4 individuals with NDPH were excluded due to not having undergone scanning, with the final sample included in the analysis consisting of 19 individuals with NDPH and 23 HCs. Demographic analysis showed that there were significant differences between the two groups in terms of GAD-7 scores, PHQ-9 scores and PSQI scores, while no significant differences were found in age and sex (see Table 2 for detailed demographic and clinical information). Compared with the healthy control group, the iron content in right paracentral, right cuneus and left precentral was significantly decreased, while the iron content in left rostral middle frontal was significantly increased after FDR correction (see Table S4). In the correlation analysis, no significant associations were observed between the QSM values of these brain regions and clinical indices. The structures, functions and QSM values of limbic system subregions and their clinical correlations In the analysis of the limbic system, four representative brain regions, namely the hypothalamus, thalamus, hippocampus and amygdala were selected to further investigate the changes in specific subregions within these areas. In the volumetric analysis of the limbic system after FDR correction, we observed significant alterations in the volumes of the thalamus and subregions of the bilateral hippocampus. Compared with the healthy control group, the volumes of the left nucleus reuniens (L-MV(Re)) in thalamus and Molecular Layer of Hippocampus head and body (molecular_layer_HP-head and molecular_layer_HP-body) in the bilateral hippocampus exhibited an increase. In contrast, other hippocampal subregions were noted to have significantly less volume, such as the Granule Cell and Molecular Layer of Dentate Gyrus head and body (GC-ML-DG-body and GC-ML-DG-head) (more details can be got in Table S7). In the correlation analysis, we identified that the volumes of the left hippocampal subregions, including left molecular_layer_HP-head, left Cornu Ammonis 1 Head (CA1-head), left GC-ML-DG-head and left parasubiculum exhibited a negative correlation with headache history(P < 0.05) (see Table S10 and Fig. 3). Fig. 3Correlation between subregions of limbic system in volume and clinical scales. A. A significant negative correlation between the volume of the Left_molecular_layer_HP-head and headache history (months). B. A significant negative correlation between the volume of the Left_CA1-head and headache history (months). C. A significant negative correlation between the volume of the Left_GC-ML-DG-head and headache history (months). D. A significant negative correlation between the volume of the Left_parasubiculum and headache history (months) Correlation between subregions of limbic system in volume and clinical scales. A. A significant negative correlation between the volume of the Left_molecular_layer_HP-head and headache history (months). B. A significant negative correlation between the volume of the Left_CA1-head and headache history (months). C. A significant negative correlation between the volume of the Left_GC-ML-DG-head and headache history (months). D. A significant negative correlation between the volume of the Left_parasubiculum and headache history (months) For the functional imaging analysis, we just observed a significant decrease in fALFF values in the right nucleus reuniens (Right MV(Re)) of thalamus (FDR < 0.05) and no significant correlation between fALFF values and clinical scales. In the ReHo analysis, Following FDR correction, a widespread diminution in ReHo values was observed across numerous subregions, which were concentrated in the amygdala (such as bilateral medial-nucleus), left hippocampus (hippocampal tail, Cornu Ammonis 3 Body (CA3-body), Cornu Ammonis 1 Body (CA1-body) and molecular_layer_HP-body), hypothalamus (right anterior superior, right tuberal superior and right anterior inferior) and thalamus (bilateral Parafascicular Nucleus (Pf), bilateral Ventral Anterior Magnocellular (VAmc) etc.). Further partial correlation analyses showed that the ReHo value of medial-nucleus in left amygdala was positively correlated with GAD-7 score(P < 0.05). The ReHo value of Right-anterior inferior in hypothalamus was positively correlated with headache history(P < 0.05). The ReHo value of Right-MV(Re) in thalamus was negatively correlated with VAS score(P < 0.05) (see Table S10 and Fig. 4). Fig. 4Correlation between subregions of limbic system in ReHo values and clinical scales. A. A significant positive correlation between ReHo values of the Medial-nucleus and GAD-7 scores. B. A significant positive correlation between ReHo values of the Right-anterior inferior and headache history (months). C. A significant negative correlation between ReHo values of the Right-MV(Re) and VAS scores Correlation between subregions of limbic system in ReHo values and clinical scales. A. A significant positive correlation between ReHo values of the Medial-nucleus and GAD-7 scores. B. A significant positive correlation between ReHo values of the Right-anterior inferior and headache history (months). C. A significant negative correlation between ReHo values of the Right-MV(Re) and VAS scores In QSM value analysis, we found that QSM values of the Central-nucleus in the left amygdala and the left parasubiculum in hippocampus were increased compared to HCs (FDR < 0.05) and no significant correlation between QSM values and clinical scales. Discussion This study represents the first exploratory investigation to assess abnormal changes in brain regions and subregions of the limbic system (including the thalamus, hippocampus, amygdala, and hypothalamus) in individuals with NDPH using 7T multimodal MRI, integrating perspectives of structure (volumetric), function (fALFF/ReHo), and iron metabolism (QSM). These observations add to the limited neuroimaging evidence base for NDPH, offering initial insights that may inform future inquiries into the neurobiological correlates of this refractory headache. Notably, some of the identified abnormalities showed associations with clinical measures (e.g., headache history and GAD-7 score). Such associations provide preliminary clues relevant to the “pain-emotion” interaction processes documented in chronic pain conditions, which may guide subsequent research focused on understanding NDPH-related neurobiological features. Whole-brain volumetric analyses in this study demonstrated significantly less volume in the right accumbens area and left cACC after FDR correction compared to HCs. The accumbens area is a core component of the reward system [42]—may relate to pain-associated motivational deficits, lowering patients’ expectations of pain relief and worsening treatment adherence. The cACC plays a crucial role in obsessive-compulsive disorder [43], social hierarchy control [44], and the regulation of emotion and pain [45, 46]. Previous migraine study had revealed a reduced volume of the cACC compared with healthy control groups [45], which contradicts our findings and further illustrates the neuroimaging differences between the two disorders. When the study focused on the limbic system (based on the priori hypothesis of high emotional comorbidity in NDPH), we observed more subtle, subregion-specific structural changes that were directly associated with clinical outcomes. In the thalamus, the volume of the L-MV(Re) (a subregion that transmits contextual information to the hippocampus [47]) was increased, which may represent a compensatory response of the brain to enhance thalamic-hippocampal communication under pain-related dysfunction. In the bilateral hippocampi, however, a mixed pattern of changes was identified: the molecular layer volume in the hippocampal head and body was increased, while the CA1-head, CA4-head and GC-ML-DG in the hippocampal head/body volume exhibited significant atrophy. Most importantly, the volume of left hippocampal subregions (molecular layer head, CA1 head, GC-ML-DG head, and para subiculum) was negatively correlated with headache history, meaning longer disease duration was associated with more severe brain region atrophy. As a core structure for encoding pain memories, hippocampal subregional atrophy may lead to abnormal consolidation of pain memories, where short-term pain stimuli are over-encoded as long-term memories, making pain difficult to resolve [21, 48–50]. Except that, previous studies have reported that the hippocampal CA4 region is involved in encoding chronic stress and negative emotional memories [51, 52]. A decrease in its ReHo may lead to disturbances in the encoding of negative emotions. In addition, negative stressful events are common precipitating factors in patients with NDPH [53], which further exacerbates the interaction between pain and emotions. Functional analyses using fALFF and ReHo further revealed widespread, FDR-corrected changes that tightly correlated with clinical symptoms. For fALFF, individuals with NDPH showed elevated activity across the frontal, parietal, temporal, and occipital lobes, as well as the cerebellum and limbic structures (bilateral amygdala, hippocampus). Cortical regions with elevated fALFF, such as inferior frontal gyrus subregions (e.g., left pars opercularis, right pars triangularis) involved in cognitive evaluation of pain [54–57] and parietal lobe which is responsible for somatosensory integration [58], reflect “hypervigilance” to pain signals. This hyperactivation may represent a compensatory mechanism attempting to enhance cortical processing to manage disordered pain signals, but long-term compensation may increase energy consumption, exacerbating neuronal exhaustion. Increasing fALFF values in the occipital and temporal lobes suggest dysfunction in visual-auditory sensory integration networks [59], potentially contributing to multimodal sensory hypersensitivity in patients [6]. Critical clinical correlations reinforced this hyperactivation’s pathological role: fALFF in the right accumbens, right hippocampus, and left lateral occipital cortex was positively correlated with GAD-7 scores and fALFF in the left transverse temporal cortex was negatively correlated with VAS pain intensity suggesting reduced excitability here may be a compensatory attempt to dampen sensory input amid severe pain. The values of fALFF in the left inferior parietal cortex and left frontal pole correlated negatively with headache history and positively with PHQ-9 scores, respectively. These associations highlight how prolonged pain disrupts sensory-cognitive processing, while executive dysfunction exacerbates mood dysregulation for individuals with NDPH. The results of Reho and fALFF complement and corroborate each other, collectively revealing functional abnormalities in specific brain regions among individuals with NDPH. It was observed that the ReHo values in multiple brain regions within the frontal lobe (left rostral middle frontal, left pars orbitals, left pars opercularies, left pars triangularis, left precentral, left caudal middle frontal and left frontal pole) and parietal lobe (left supramarginal, left inferior parietal, left superior parietal and left postcentral) exhibited a significant increase. Consistent with the findings of fALFF, this observation further reflects the enhanced abnormal activation and elevated synchrony in the aforementioned brain regions. In addition, the ReHo values in the right pallidum, bilateral parahippocampal gyrus, and other relevant brain regions exhibited a significant reduction. As reported in previous studies, the pallidum not only participates in pain regulation but also is involved in the mechanism of comorbid depression [60], which may serve as one of the targets for addressing the high comorbidity of emotional disorders in NDPH. In the correlation analysis, the ReHo values of the left parahippocampal, left superior parietal, left inferior parietal, left supramarginal, and left parsopercularies were negatively correlated with headache history. The ReHo value of the left frontal pole was positively correlated with PHQ-9 scores. We speculate that the observed functional abnormalities in cortical and subcortical structures may disrupt cross-regional signal transmission and weaken descending inhibitory pathways for pain regulation, which could also contribute to treatment resistance in NDPH. These results indicate that NDPH is not confined to a single brain region but involves coordinated damage across multiple networks, including sensory integration, reward processing, and signal transmission, providing a macroscopic perspective for understanding its “persistent pain-dysfunction” cycle. When focusing on the subregional level of the limbic system, it was observed that the ReHo values of multiple substructures were significantly reduced. This change echoes the functional abnormalities at the macroscopic level, refines the macroscopic network impairment to the subregional level, and further reveals the pathological feature of weakened synchronous activity of neuronal clusters within the limbic system. This exploration also holds breakthrough significance for revealing the pathological mechanisms of NDPH via QSM analysis. Iron is involved in important physiological processes such as mitochondrial respiratory chain function [61], myelin synthesis [62], and neurotransmitter metabolism [63, 64]. Its metabolic imbalance is closely related to neuronal damage, oxidative stress, and neuroinflammation [65]. In headache research, abnormal iron deposition patterns have also been reported in patients with medication overuse headache and migraine [14, 66].However, as a type of chronic headache with unknown etiology and poor treatment response, the status of brain iron metabolism in NDPH has never been systematically studied, making the QSM results of this study a key evidence to fill the gap in the field. After FDR correction, compared with the healthy control group, the iron content in the right paracentral lobule, right cuneus, and left precentral gyrus was significantly reduced, whereas the iron content in the left rostral middle frontal gyrus was significantly elevated. Within the limbic system, elevated QSM values were observed in the central nucleus of the left amygdala and the left parasubiculum of the hippocampus when compared to HCs (FDR < 0.05). These abnormal patterns of iron deposition provide new potential targets for the treatment of NDPH. Given that iron metabolism modulators (such as the iron chelator deferiprone) have shown potential in reducing brain iron deposition and improving neurological function in clinical trials of Alzheimer’s disease [67], this study suggests that targeted intervention for abnormal iron metabolism in specific brain regions of individuals with NDPH may break the vicious cycle of “iron overload-oxidative stress-neuronal damage-pain exacerbation”. Although the current results need to be verified with larger samples (only 19 individuals with NDPH enrolled), this exploration is the first to advance the research on the pathological mechanism of NDPH from the “structural-functional connectivity” level to the “molecular metabolism” level, providing a new perspective for understanding its “refractory” characteristics. There is currently no specific treatment plan for NDPH, and most headache specialists use the acute and preventive medications they employ for chronic migraine to treat NDPH [68–70].While the “abnormal iron metabolism” suggested by this study may be one of the potential reasons for the poor response of NDPH to conventional treatments, which also lays a theoretical foundation for the development of a combined treatment strategy of “iron metabolism regulation and pain management”. We must acknowledge that our study has certain limitations. First, the cross-sectional design precludes the establishment of causal relationships; longitudinal studies tracking NDPH from its onset are therefore needed to clarify the temporal association between neural abnormalities and pain chronicity. Second, given that NDPH is a rare disorder, the sample size of this study was relatively small—larger-scale multicenter studies are thus required in the future to validate the findings. Third, potential alternative explanations for the findings include prior medication use (though patients had discontinued ineffective treatments for ≥ 2 weeks before scanning) and comorbidities. Although we included GAD-7 scores, PHQ-9 scores, and PSQI scores as covariates to further control for potential confounders, the interpretation of the results may still be biased due to the small sample size. Finally, constrained by the sample size, this study treats NDPH as a homogeneous group; future studies with larger cohorts should stratify patients by phenotype (e.g., medication-responsive vs. refractory subtypes) to clarify subgroup-specific neurobiological changes. However, this study holds significant clinical implications: it focused on the abnormal patterns of structure-function-iron metabolism in the limbic system of individuals with NDPH, providing potential insights for further understanding the “pain-emotion” interaction. The correlations between relevant indicators and clinical variables offer potential quantifiable prognostic markers, while abnormalities in iron metabolism suggest that targeted iron regulation may serve as a novel therapeutic strategy. Future research should expand the sample size, adopt a longitudinal design to track the dynamic relationship between neural abnormalities and disease progression, combine QSM with cerebrospinal fluid iron detection to link brain iron levels with systemic metabolism, and conduct clinical trials of iron-targeted therapies to evaluate their effects on pain and emotional symptoms. However, the high cost and relatively low accessibility of 7T MRI, coupled with the small sample size included in this study, have posed certain challenges to the clinical translation of our findings. Conclusion Using 7T multimodal MRI, this study systematically revealed structural, functional, and iron metabolism abnormalities in NDPH from whole-brain networks to limbic subregions. Widespread whole-brain damage and limbic subregional changes collectively form a “pain-emotion-metabolism” pathological cycle, providing preliminary insights into potential mechanisms of chronicity, with findings requiring replication. These findings provide potential insights for the future development of novel therapeutic approaches and the screening of potential biomarkers. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3 Supplementary Material 4 Supplementary Material 4 Supplementary Material 5 Supplementary Material 5 Supplementary Material 6 Supplementary Material 6 Supplementary Material 7 Supplementary Material 7 Supplementary Material 8 Supplementary Material 8 Supplementary Material 9 Supplementary Material 9 Supplementary Material 10 Supplementary Material 10