Altered spontaneous brain activity is associated with cognitive impairment in patients with Kallmann syndrome Authors: Yan Fu, Xueying Wang, Feifei Wu, Jinfan Zhang, Han Wu, Yuanhao Li, Yafang He, Meichao Men, Wenzhen Zhu, Jiada Li, Alessandro Grecucci, Yuanchao Zhang, Xiaoping Yi, Bihong T Chen Published: 2026 ABSTRACT AbstractObjectivesThis study aimed to investigate alterations in spontaneous brain activity in Kallmann syndrome (KS) and their associations with cognitive impairment and genetic mutations.MethodsWe prospectively enrolled 100 patients with KS and 100 matched healthy controls. All participants underwent brain MRI and neuropsychological assessment targeting global cognition (Montreal-Cognitive-Assessment, MoCA), executive function (Stroop-Color-and-Word-Test, SCWT), cognitive flexibility (Trail-Making-Test), and working memory (Digit-Span-Test, DST; Visual-Reproduction, VR). Regional brain activity was quantified using the amplitude of low-frequency fluctuation (ALFF) metric. We further conducted brain-cognition correlations, genetic association analysis with whole-exome sequencing, and extrapolated neuroimaging-transcription-association analysis.ResultsCompared to the controls, patients with KS exhibited significant cognitive deficits across most domains, with 36% reached the clinical cutoff for cognitive impairment. Patients showed increased ALFF predominantly in the fronto-insular executive network and salience network, but decreased ALFF in the cerebellum and left angular gyrus (cluster-level GRF-corrected-p < 0.05). The increased ALFF in fronto-insular regions was positively correlated with MoCA (r = 0.215, FDR-p = 0.037) and SCWT-B (r = 0.253, FDR-p = 0.026), while the reduced ALFF in the left angular gyrus was negatively correlated with VR and DST-forward (r=-0.270, FDR-p = 0.011; r=-0.280, FDR-p = 0.011, respectively). Exploratory genetic analysis revealed that a higher cumulative mutation burden was associated with lower ALFF in cerebellum and angular gyrus, while ANOS1 and PROKR2 mutations mapped to distinct network alterations. Furthermore, the spatial pattern of ALFF alterations was significantly associated with extrapolated gene expression profiles enriched for neurodevelopment and cell migration.ConclusionsThis study found a genetically related trade-off between compensatory executive hyperactivation and dysfunctional brain activity with suppression as a potential neuroimaging biomarker for cognitive impairment risk in patients with KS.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12958-026-01554-9.Keywords: Kallmann syndrome, Cognitive impairment, Magnetic resonance imaging, Amplitude of low-frequency fluctuation, Neuroimaging-transcription Introduction Kallmann syndrome (KS) is a rare genetic disorder defined by the co-occurrence of congenital hypogonadotropic hypogonadism and impaired olfaction with anosmia or hyposmia [1, 2]. It is a neurodevelopmental disorder stemming from the failed embryonic migration of gonadotropin-releasing hormone (GnRH) neurons from the olfactory placode to hypothalamus [3, 4]. Genetic mutations in key regulators such as ANOS1 [5], HS6ST1 [6, 7], and NHLH2 [8] have been implicated in fundamental cellular processes like cell migration, axonal guidance, [and neuronal differentiation, which may have broad-ranging effects on central nervous system development in patients with KS. Patients frequently exhibit impairments in cognition [9–11], while the neural correlates underlying the cognitive and behavioral manifestations of patients with KS remain largely unknown. Resting-state functional magnetic resonance imaging (rs-fMRI) offers a valuable, non-invasive tool for assessing intrinsic brain activity [12, 13]. Among the rs-fMRI metrics, the amplitude of low-frequency fluctuation (ALFF) is particularly useful as it quantifies the intensity of spontaneous neuronal activity in the blood-oxygen-level-dependent (BOLD) signal within a specific frequency band (typically 0.01–0.08 Hz) [14]. Alterations in ALFF have been identified as a sensitive biomarker for dysfunctional regional brain activity in autism and other psychiatric disorders [15, 16], and have been found to be associated with cognition in autism and Alzheimer’s disease [17, 18]. Despite the broad application of rs-fMRI in various disorders, research investigating spontaneous brain activity in patients with KS is limited. Among the few published the functional MRI studies, one study reported significantly reduced functional connectivity within the frontal-limbic system in patients with idiopathic hypogonadotropic hypogonadism 19, while another study employing both task-based and rs-fMRI reported functional reorganization in motor-related regions associated with mirror movements in patients with KS [20]. These studies supported the notion of widespread neural circuit abnormalities associated with this disorder. In addition, most published neuroimaging studies on patients with KS have focused on structural abnormalities in olfaction-related brain regions, with no assessment into intrinsic brain activity, and some of the studies have been conducted on small and heterogenous samples [21, 22]. To date, no study has systematically characterized spontaneous neural activity in patients with KS using sensitive metrics such as ALFF for detection of brain regional functional abnormalities. To bridge this gap, we performed a prospective, multi-modal study in a large cohort of patients with KS and age-sex-matched healthy control (HC) participants. We utilized ALFF to characterize whole-brain intrinsic activity and investigated its correlation with cognitive performance across multiple domains. In addition, we performed an exploratory genotype-phenotype analysis to associate common genetic mutations with specific brain and cognitive outcomes. Furthermore, by integrating regional ALFF patterns with extrapolated spatial gene expression data from the Allen Human Brain Atlas [23, 24], we aimed to identify the potential biological pathways and molecular drivers underlying these functional alterations. We hypothesized that patients with KS would exhibit aberrant intrinsic activity in large-scale networks involved in cognition and emotion, which would correlate with functional deficits and reflect disruptions in gene pathways critical for neurodevelopment. Materials and methods Participants and clinical assessment Patients with KS were prospectively recruited from the outpatient clinics of endocrinology and metabolism in Xiangya Hospital between July 2023 and March 2025. The HC participants matched by age and sex were recruited from local community. The detailed recruitment process for this cohort is presented in Fig. 1. All patients included in this study fulfilled the diagnostic criteria for KS, established through clinical/hormonal evaluation of hypogonadotropic hypogonadism and confirmed olfactory dysfunction using the University of Pennsylvania Smell Identification Test [25, 26]. All participants underwent a structured clinical interview to assess whether they had any neurological disorders such as Alzheimer’s disease, dementia or psychiatric disorders by the clinical team. Flow-chart illustrating the enrollment process for the patients with Kallmann syndrome (KS) and the healthy controls Beyond diagnostic confirmation, a comprehensive clinical, genetic and hormonal profiling was conducted for all patients. The whole-exome sequencing was performed to characterize the genetic background. Hormonal status was assessed by measuring serum levels of follicle-stimulating hormone, luteinizing hormone, testosterone, estradiol, progesterone, and prolactin, collected both at the time of initial diagnosis and at the time of rs-fMRI scanning. Detailed clinical history was obtained to assess fetal sex steroid deficiency and pubertal development, specifically recording the presence of cryptorchidism and micropenis in male patients. Furthermore, treatment history was systematically categorized by modality and duration to control for exogenous hormonal effects. Cognitive and psychological assessment All psychometric instruments used in this study were validated in Chinese language. The cognitive assessment for this study included the Montreal Cognitive Assessment (MoCA) for global cognitive function [27], with the total score ranging from 0 to 30, with a score < 26 indicating mild cognitive impairment [28, 29]. The Stroop Color and Word Test (SCWT) was employed to evaluate selective attention and response inhibition [30]. It comprised three subtasks: SCWT-A (reading color words) and SCWT-B (naming ink colors) assessed processing speed and attention; SCWT-C (naming ink colors of incongruent words) specifically measured impulse control and the ability to inhibit cognitive interference. The Digit Span Tests (DST) [31, 32] included the DST-A (assigned number forwards) for measuring attention and immediate memory, and the DST-B (assigned number backwards) for assessing working memory and executive control. The Trail Making Test (TMT) also consisted of two parts [33]. TMT-A (connecting numbers sequentially) measured visual scanning and processing speed. TMT-B (alternating between numbers and letters) assessed cognitive flexibility and executive shifting. The outcome metric was time to completion (seconds), with longer durations indicating poorer performance. The Visual Reproduction (VR) subtest of the Rey-Osterrieth Complex Figure Test was used to assess visuospatial memory and constructional ability [34]. All clinical interviews, cognitive assessment and hormonal evaluations were conducted on the same day as the brain MRI scan. MRI data acquisition All brain MRI data were acquired on the same Siemens MAGNETOM Prisma 3.0 Tesla scanner with a 64-channel head coil. T1-weighted 3D Magnetization Prepared - Rapid Gradient Echo structural scans were acquired in sagittal orientation using the following parameters: voxel size of 1.0 × 1.0 × 1.0 mm3, acquisition matrix 256 × 256 mm2, repetition time [TR] 2300 milliseconds (ms), echo time [TE] 3.0 ms, flip angle = 9°, 208 slices with no gap. The rs-fMRI data were acquired with an echoplanar imaging sequence (TR/TE = 2000/34 ms; 66° flip angle; 75 slices with no gap; 2 mm slice thickness; 210 × 210 mm FOV; 104 × 104 matrix). Additional imaging data such as the T2-weighted sequence and fluid attenuation inversion recovery sequence were obtained for evaluation of incidental brain abnormalities. During the scan, participants were instructed to rest with their eyes closed. All scans were visually inspected for quality and incidental findings. ALFF analysis The rs-fMRI data were preprocessed using the Data Processing and Analysis for Brain Imaging (DPABI; http://rfmri.org/DPABI) toolbox [35] in MATLAB R2021b. The first 10 volumes of each run were discarded to allow for magnetic field stabilization. The remaining images were corrected for slice acquisition timing and realigned to account for head motion. Individual T1-weighted structural images were co-registered to the mean functional image and segmented into tissue classes. Nonlinear spatial normalization to Montreal Neurological Institute (MNI) space was performed using the DARTEL approach to create a study-specific template. Functional images were then normalized to the template and resampled to an isotropic voxel size of 3 × 3 × 3 mm³. Spatial smoothing was performed using a Gaussian kernel with a full width at half maximum (FWHM) of 6 mm. Temporal band-pass filtering (0.01–0.10 Hz) was applied to minimize low-frequency drift and high-frequency physiological noise. Nuisance regression was performed including white matter, cerebrospinal fluid signal, and Friston 24-parameter head motion covariates. ALFF was calculated to quantify regional spontaneous neural activity. For each voxel, the preprocessed time series was transformed into the frequency domain using fast Fourier transform (FFT). The power spectrum was obtained, and the square root of the power was averaged within the 0.01–0.10 Hz frequency band to derive the ALFF value. To reduce global effects and inter-individual variability, the resulting ALFF maps were standardized using voxel-wise z-transformation prior to group-level statistical analyses. Group differences in ALFF maps were examined using voxel-wise two-sample t-tests implemented in DPABI, with age and sex included as covariates. Meanwhile, voxel-wise T-statistic maps were yielded for neuroimaging - extrapolated transcription association analysis. Brain-cognition correlation analysis Partial correlation analyses were performed between the mean ALFF values from the significant clusters identified and the raw scores from neurocognitive tests, controlled for age, gender, years of education, treatment modality, and treatment duration to rule out potential confounders. Model-based mediation analysis was performed using PROCESS macro for SPSS (version 3.5), adjusting for relevant confounders. Specifically, the simple mediation model (Model 4 in PROCESS) was used to test for potential mediation effects. Genotype-phenotype correlation analysis To investigate the impact of genetic heterogeneity within the KS cohort, an exploratory genotype-phenotype association analysis was performed. All patients with KS in this study underwent genetic testing using whole-exome sequencing (WES) as reported previously [36]. Non-synonymous rare sequencing variants (RSVs) with a minor allele frequency < 1% in the Genome Aggregation Database, Chinese Millionome Database, dbSNP, ExAC and 1000 Genomes in IHH-associated genes were screened [37, 38]. We focused on the six most frequently mutated genes in our cohort: CHD7, CCDC141, PROKR2, FGFR1, OTUD4, and ANOS1. We then systematically assessed the associations between the presence of mutations in these genes, both individually and as a cumulative burden, and our key clinical and neuroimaging phenotypes. For each patient, binary indicators (1 for presence, 0 for absence) were created for mutations in each of these six genes. Additionally, two summary variables were generated: ‘Mut_or Not’ (a binary indicator for the presence of a mutation in any of the six genes) and ‘Mut_Number’ (a quantitative count of mutated genes from the list for each patient). We then examined the associations of these eight genetic indicators with the cognitive scores or with the mean ALFF from the five significant clusters. For binary genetic variables (genes mutant vs. wild-type and any mutation vs. no mutation), associations were assessed using the Wilcoxon rank-sum test. Fold change (log2FC) between groups was calculated to indicate the direction and magnitude of differences. For the mutation burden (‘Mut_Number’ variable), a linear regression model (lm function in R) was used to assess its relationship with cognitive score and ALFF value, with the estimated regression coefficient (Coef) representing the strength and direction of the relationship. A p-value < 0.05 was considered statistically significant for this exploratory analysis. Given the hypothesis-generating nature of this specific genotype-phenotype assessment, we intentionally did not apply multiple comparison corrections to prevent discarding potential early signals in this rare cohort. Neuroimaging–tracription association analysisns To investigate potential biological pathways underlying the observed regional ALFF alterations, we performed a neuroimaging-transcription association analysis, linking regional ALFF differences quantified across the 116 AAL regions, to spatial patterns of extrapolated gene expression from the Allen Human Brain Atlas. Detailed methodological steps were described as follows. First, regional ALFF deviations were quantified across the 116 AAL brain regions (http://surfer.nmr.mgh.harvard.edu/). For each region, the mean T-statistic value (reflecting group differences in ALFF between the KS group and the HC group) was extracted from the vertex-wise group comparison. In parallel, gene expression data from the Allen Human Brain Atlas (http://human.brain-map.org/) was mapped onto the same 116 regions using the abagen toolbox (https://github.com/rmarkello/abagen), a standardized framework for processing and aggregating Allen Human Brain Atlas data. This process yielded a regional gene expression matrix comprising expression levels for 15,633 genes across 116 regions. Partial least squares (PLS) regression was then applied to examine the relationship between gene expression and ALFF deviations, with regional transcription data as predictors and the ALFF T-values as the response variable. In the neuroimaging-transcriptomic analysis, the first component (PLS1) represented the set of gene expression weights that best explained the spatial variation in neuroimaging data. The statistical significance of PLS1 was assessed using a non-parametric permutation test with 1000 iterations. To estimate the variability of each gene’s weight, we performed bootstrapping by resampling the 116 brain regions 1000 times with replacement. A Z-score for each gene was then calculated by dividing its weight by the corresponding bootstrap-derived standard deviation. Genes with Z-scores ≥ 4 or ≤–4 were identified and subjected to Gene Ontology (GO) and KEGG pathway enrichment analysis using the Metascape platform (https://metascape.org/). Statistical analysis All statistical analyses were conducted using SPSS (version 26.0), MATLAB R2021b and R software (version 4.1.1). Data normality was assessed using the Shapiro-Wilk test. Continuous variables are reported as mean ± standard deviation (SD) for normally distributed data, or median (interquartile range, IQR) for non-normally distributed data. Categorical variables are expressed as frequencies and percentages. Demographic and clinical differences between groups were evaluated using Student’s t-tests, Mann-Whitney U tests, or chi-square (χ²) tests, as appropriate. These tests of significance were two-tailed, with p < 0.05 being considered statistically significant. Between-group contrasts for ALFF were initially thresholded at a conservative voxel-wise significance level of p < 0.001 (uncorrected), and subsequent multiple-comparison correction was performed at the cluster level using Gaussian random field (GRF) theory. Results were considered statistically significant at a GRF-corrected p < 0.05. The False Discovery Rate (FDR) was applied using the Benjamini-Hochberg procedure for the brain-cognition partial correlations. For these analyses, significance was defined as an adjusted p-value (p-FDR) < 0.05. Results Demographic data and clinical assessment A total of 100 patients with KS and 100 age-sex-matched HC participants were enrolled into the study. Demographic and cognitive assessment data for all participants are summarized in Table 1. There were no significant differences between the two groups in age (p = 0.201), sex (p = 0.800), or years of education (p = 0.159). Table 1Demographic and clinical assessment of patients with Kallmann syndrome (KS) and the healthy controlsVariablesPatients with KS (n = 100)Healthy controls (n = 100)p valueAge (years), mean ± SD28.6 ± 7.630.0 ± 8.40.201Male gender, n (%)92 (92%)91 (91%)0.800Education (years)15.0 (12.0, 16.0)15.0 (12.0, 16.0)0.159MoCA, mean ± SD26.2 ± 2.728.0 ± 1.6< 0.001***Cognitive impairment#, n (%)36 (36%)0 (0%)< 0.001***SCWT-A100 (87,100)95 (90, 100)0.952SCWT-B67 (60, 78)66 (58, 72)0.167SCWT-C37 (30, 43)40 (35, 46)0.003**DST-Forward9 (8, 10)10 (9, 11)0.020*DST-Backward5 (5, 6)6 (5, 7)< 0.001***TMT_A32.0 (25.3, 37.0)32.0 (25.0, 40.8)0.689TMT_B80.0 (68.0, 105.8)67.0 (53.0, 81.5)< 0.001***Visual Reproduction11 (10, 12)12 (10, 13)0.031*Results are expressed as frequencies for the categorical variables, mean ± standard deviation (SD) for normally distributed continuous variables and median (interquartile range (IQR)) for continuous variables that do not fit a normal distribution. DST, Digit Span Test; MoCA, Montreal Cognitive Assessment; SCWT, Stroop Color and Word Test; TMT, Trail Making Test#Montreal Cognitive Assessment total score < 26, indicating cognitive impairment*, p < 0.05, **, p < 0.01, ***, p < 0.001 Demographic and clinical assessment of patients with Kallmann syndrome (KS) and the healthy controls Results are expressed as frequencies for the categorical variables, mean ± standard deviation (SD) for normally distributed continuous variables and median (interquartile range (IQR)) for continuous variables that do not fit a normal distribution. DST, Digit Span Test; MoCA, Montreal Cognitive Assessment; SCWT, Stroop Color and Word Test; TMT, Trail Making Test #Montreal Cognitive Assessment total score < 26, indicating cognitive impairment *, p < 0.05, **, p < 0.01, ***, p < 0.001 Patients with KS showed significantly decreased cognitive testing scores across multiple domains including global cognition, inhibitory control, working memory, cognitive flexibility, and executive function as compared to the HC groups. Specifically, compared to the HC group, patients with KS exhibited significantly lower MoCA (26.2 ± 2.7 vs. 28.0 ± 1.6, p < 0.001), VR (median = 11.0 vs. 12.0, p = 0.031), DST-forward (median = 9.0 vs. 10.0, p = 0.020), DST-backward (median = 5.0 vs. 6.0, p < 0.001), SCWT-C (median = 37.0 vs. 40.0, p = 0.003), and longer TMT-B (median = 80.0 s vs. 67.0 s, p < 0.001), with 36% of this group scoring below the MoCA cut-off of 26 [28, 29]. Table 2 presents the detailed clinical profile of the KS cohort. The cohort consisted of 92 males and 8 females. Genetically, the majority (63%) harbored monogenic mutations, while 20% had oligogenic variants. Detailed variant-level genetic information for each patient, including gene name, variant notation (cDNA/protein), zygosity, and variant classification, is provided in Supplementary Table S1. Phenotypically, all patients presented with absent puberty and typical signs of hypogonadotropic hypogonadism. Among males, micropenis was universal (100%), and cryptorchidism was common (32.6%). At the time of the brain MRI scan, 90% of patients were receiving hormone replacement therapy, mostly for 1–5 years. Table 2Detailed clinical and hormonal characteristics of patients with Kallmann syndrome Characteristics Total (n = 100)Males (n = 92)Females (n = 8)Demographics Age at MRI (years), mean ± SD28.6 ± 7.628.6 ± 7.631.5 (26.3, 37.3) Age at diagnosis (years), mean ± SD20.5 ± 6.620.5 ± 6.617.0 (15.0, 19.5)Developmental & Pubertal History History of Cryptorchidism, n (%)N/A30 (32.61%)N/A History of Micropenis, n (%)N/A92 (100%)N/A Testicular Volume (mL, bilateral)N/A2.30 (1.05, 4.00)N/A Absent Puberty, n (%)100 (100%)100 (100%)100 (100%)Genetic mutation status Monogenic63 (63%)58 (63.04%)5 (62.50%) Oligogenic20 (20%)18 (19.57%)2 (25%) No Mutation Identified17 (17%)16 (17.39%)1 (12.50%)Hormonal Profile (at diagnosis) Follicle-Stimulating Hormone (IU/L)0.67 (0.22, 1.08)0.70 (0.23, 1.09)0.23 (0.04, 0.88) Luteinizing Hormone (IU/L)0.34 (0.11, 0.80)0.40 (0.13, 0.85)0.11 (0.02, 0.20) Prolactin (ng/ml)7.50 (4.83, 10.90)7.50 (4.83, 10.90)7.83 (4.00, 15.23) Estradiol (pg/mL)17.00 (11.00, 23.45)17.18 (11.00, 23.60)16.50 (5.15, 19.15) Progesterone (ng/ml)0.55 (0.18, 0.80)0.54 (0.18, 0.80)0.90 (0.18, 1.88) Testosterone (ng/mL)0.11 (0.02, 0.34)0.10 (0.02, 0.34)0.17 (0.04, 0.54)Hormonal Profile (at MRI time point) Follicle-Stimulating Hormone (IU/L)1.07 (0.45, 2.12)1.25 (0.51, 2.14)0.31 (0.10, 1.64) Luteinizing Hormone (IU/L)0.39 (0.10, 1.37)0.40 (0.17, 1.40)0.10 (0.02, 0.78) Prolactin (ng/ml)7.91 (6.40, 11.50)7.96 (6.40, 11.07)6.94 (5.40, 17.04) Estradiol (pg/mL)15.00 (10.83, 21.03)15.00 (10.80, 20.43)19.75 (11.25, 23.50) Progesterone (ng/ml)0.56 (0.20, 0.85)0.55 (0.20, 0.83)0.62 (0.21, 1.25) Testosterone (ng/mL)2.09 (0.90, 4.48)2.85 (1.01, 4.80)0.30 (0.14, 0.67)Treatment details Treated at MRI, n (%)90 (90%)83 (90.22%)7 (87.50%)Treatment Modality, n (%) Null10 (10%)9 (9.78%)1 (12.50%) Sex Steroid Replacement40 (40%)39 (42.39%)1 (12.50%) Gonadotropin Therapy37 (37%)32 (34.78%)5 (62.50%) GnRH Pump Therapy13 (13%)12 (13.04%)1 (12.50%)Treatment Duration (years) 0 years10 (10%)9 (9.78%)1 (12.50%) 0 ~ 1 years20 (20%)20 (21.74%)3 (37.50%) 1 ~ 5 years47 (47%)44 (47.83%)3 (37.50%) 5 ~ 10 years13 (13%)10 (10.87%)1 (12.50%) > 10 years10 (10%)9 (9.78%)0Results are expressed as frequencies for the categorical variables, mean ± standard deviation (SD) for normally distributed continuous variables and median (interquartile range (IQR)) for continuous variables that do not fit a normal distribution Detailed clinical and hormonal characteristics of patients with Kallmann syndrome Results are expressed as frequencies for the categorical variables, mean ± standard deviation (SD) for normally distributed continuous variables and median (interquartile range (IQR)) for continuous variables that do not fit a normal distribution ALFF alterations Whole-brain analysis revealed significant alterations in ALFF in patients with KS as compared to the HC participants (Fig. 2). The patients with KS showed markedly increased ALFF in core nodes of the salience network, default mode network, and fronto-limbic circuits. These hyperactive regions were primarily localized to the bilateral insula as well as widespread areas of the left frontal and temporal cortex. The clusters with significant increased ALFF included the following: (1) Cluster 2 encompassing the left middle temporal gyrus, superior temporal gyrus, inferior temporal gyrus, and limbic lobe, Cluster 3 encompassing the left inferior, middle, and superior frontal gyri, insula, and limbic lobe, Cluster 4 encompassing the right insula, left inferior frontal gyrus, and right rolandic operculum. Conversely, significantly decreased ALFF was observed in the cerebellum and the left angular gyrus, such as Cluster 1 encompassing the left cerebellum posterior lobe and Cluster 5 encompassing the left angular gyrus (Peak MNI coordinates and cluster sizes are detailed in Table 3). Brain regions with increased (highlighted in orange color) or decreased amplitude of low-frequency fluctuations (ALFF) (highlighted in blue color) in the patients with Kallmann syndrome as compared to the healthy controls Table 3Brain regions with altered amplitude of low-frequency fluctuations (ALFF) in the patients with Kallmann syndrome (KS) as compared to the healthy control (HC) participantsClusterMNI peak coordinatesBrain regions involvedXYZT valueCluster sizeResults1Cerebellum Posterior_L−42−66−544.818332HC > KS2Middle Temporal_L−3612−42−5.981307HC < KSSuperior Temporal_LInferior Temporal_LLimbic Lobe_L3Inferior Frontal_L−5736−7.1272087HC < KSMiddle Frontal_LSuperior Frontal_LInsula_LLimbic Lobe_L4Insula_R39315−4.536144HC < KSInferior Frontal_LRolandic operculum_R5Angular_Gyrus_L−57−63364.749110HC > KS6Frontal Lobe_R31248−5.047230HC < KSLimbic Lobe_RCingulate Gyrus_RVoxel-level significance of p < 0.001 (uncorrected), and subsequent multiple-comparison correction was performed at the cluster level using Gaussian random field (GRF). Cluster-level significance of GRF-corrected p < 0.05. Abbreviations: MNI, Montreal Neurological Institute template; L, left; R, right Brain regions with altered amplitude of low-frequency fluctuations (ALFF) in the patients with Kallmann syndrome (KS) as compared to the healthy control (HC) participants Voxel-level significance of p < 0.001 (uncorrected), and subsequent multiple-comparison correction was performed at the cluster level using Gaussian random field (GRF). Cluster-level significance of GRF-corrected p < 0.05. Abbreviations: MNI, Montreal Neurological Institute template; L, left; R, right Associations between ALFF and cognitive scores Significant correlations were noted between the altered ALFF and cognitive scores (Fig. 3 and Fig. 3a). Specifically, there were significant positive correlations of the ALFF of Cluster 3 with both MoCA score (r = 0.215, FDR-corrected p = 0.037) (Fig. 3b) and SCWT-B score (r = 0.253, FDR-corrected p = 0.026) (Fig. 3c). The ALFF of Cluster 5 was negatively correlated with both VR score (r = −0.270, FDR-corrected p = 0.011) (Fig. 3d) and DST-A scores (r = −0.280, FDR-corrected p = 0.011) (Fig. 3e). Correlation between the amplitude of low-frequency fluctuations (ALFF) values and the cognitive testing scores in the patients with Kallmann syndrome. Note: (a) Brain map image showing the significant brain regions/clusters with altered ALFF in patients with Kallmann syndrome as compared to the healthy controls. b Positive correlation between the ALFF value of brain cluster 3 and the Montreal Cognitive Assessment (MoCA) score (r = 0.215, FDR-corrected p = 0.037). c Positive correlation between the ALFF value of brain cluster 3 and the Stroop Color and Word Test-(SCWT)-B score (r = 0.253, FDR-corrected p = 0.026). d Negative correlation between the ALFF value of brain cluster 5 and the Visual Reproduction score (r = −0.270, FDR-corrected p = 0.011). e Negative correlation between the ALFF value of brain cluster 5 and the Digit Span Test- (DST)-forward score (r = −0.280, FDR-corrected p = 0.011). f Mediation analysis data showing the Visual Reproduction score significantly mediating the relationship between the ALFF of Cluster 5 and DST-forward score and accounting for 24.39% of the total effect after adjusting for age, sex, and years of education. The mediation analysis showed that VR performance significantly mediated the relationship between the ALFF of Cluster 5 (left angular gyrus) and DST-forward (Fig. 3f). The indirect effect of the ALFF of Cluster 5 on DST-forward performance via VR was significant (β = −0.40, 95% CI [−0.89, −0.07]), accounting for 24.39% of the total effect after adjusting for age, sex, and years of education. Genotype-phenotype association Exploratory analysis of genotype-phenotype correlation in patients with KS identified several associations (Fig. 4). Specifically, the presence of an OTUD4 mutation was significantly associated with better performance (longer completion time) on the TMT-A (log2FC = −0.445, p = 0.038). Additionally, the presence of an FGFR1 mutation was associated with better performance on the VR test (log2FC = 0.263, p < 0.050). The ANOS1 mutation was significantly associated with lower ALFF value in cluster 1 (log2FC = −0.944, p = 0.010), and the presence of an PROKR2 mutation was significantly associated with lower ALFF value in cluster 5 (log2FC = −1.105, p = 0.029) as denoted with the largest blue dot on Fig. 4. Finally, the higher Mut_Number was associated with higher ALFF value of cluster 4 (Coef = 0.080, p = 0.003). No other significant associations were found between the analyzed genetic variants and other cognitive scores, nor with the mean ALFF values of other brain clusters identified in the primary analysis (all p > 0.05). Bubble plot summarizing the associations of the common gene mutations with cognitive testing scores from the Trail Making Test-A and Visual Reproduction task, and brain clusters showing altered amplitude of low-frequency fluctuations (ALFF) including clusters 1, 4 & 5 in the patients with Kallmann syndrome. Note: The top and bottom rows show the common gene mutations and mutation burden. For categorical comparisons (presence vs. absence of mutations in specific genes, or any mutation vs. none), bubble color represents the log2 fold change (log2FC) of the score in the mutant group relative to the wild-type group. Red oval dots indicate positive correlations and blue oval dots indicate negative correlations. For the continuous variable analysis (mutation number), the green and purple gradient color indicates the regression coefficient (Coef) from the linear model. In all panels, bubble size is proportional to the statistical significance, represented as -log10(P-value). For example, the largest red dot shows the presence of an FGFR1 mutation being significantly associated with higher VR score (log2FC = 0.263, p < 0.050). The deep green dot shows mutation number being associated with higher ALFF value in cluster 4 (Coef = 0.080, p = 0.003). *, p < 0.05, **, p < 0.01. Neuroimaging–transcription association data The neuroimaging-transcription association analysis revealed a robust link between regional ALFF deviations and gene expression profiles. This component accounted for 23.34% of the variance of the statistical contrast (permuted p < 0.0001). Applying a stringent threshold of Z-score > |4|, we identified 3,999 contributing genes. Genes with negative associations (Z < −4, n = 2,084), including the canonical KS gene ANOS1, were linked to regions with decreased ALFF. Conversely, Genes positively associated with the ALFF alteration pattern (Z > 4, n = 1,915), including WDR11, were linked to brain regions with increased ALFF in patients with KS. Functional enrichment analysis revealed distinct biological profiles for these two gene sets. For genes with Z < − 4, significantly enriched Gene Ontology (GO) terms included processes related to neurodevelopment, organogenesis, and neuronal migration, such as ‘sensory organ development’ (GO:0007423), ‘brain development’ (GO:0007420), and ‘positive regulation of cell migration’ (GO:0030335) (Fig. 5a). In contrast, the gene set with Z > 4, was enriched for pathways associated with transcriptional and translational regulation. These included ‘mRNA metabolic process’ (GO:0016071), ‘DNA-templated transcription’ (GO:0006351), ‘regulation of translation’ (GO:0006417), and ‘protein localization to organelle’ (GO:0033365) (Fig. 5b). The statistical significance was set as with a threshold of FDR corrected p < 0.05. Functional enrichment analysis of the pivotal genes associated with amplitude of low-frequency fluctuations (ALFF) alterations in patients with Kallmann syndrome. Note: The plot displays the most significantly enriched biological pathways for the gene set correlated with decreased ALFF (a) and increased ALFF (b). The color scale shows the enrichment significance as the log10-transformed, false discovery rate (FDR) corrected p-value (Log10(p-value)). The red circles show pathways with lower statistical significance, and the green circles show pathways with higher statistical significance. The size of the circles shows the gene count. Discussion This multimodal study showed a distinct pattern of altered intrinsic brain activity as measured by ALFF, which was correlated with cognitive impairment reflected by measures of global cognition, inhibitory control, working memory, cognitive flexibility, and executive function, and was linked to a specific neurodevelopmental gene expression profile in patients with KS. The altered ALFF was not restricted to olfactory or hypothalamic dysfunction, but rather involved an extensive cognitive network. In addition, specific genetic variants, including mutations in canonical KS genes such as ANOS1 and PROKR2, were associated with brain activity and cognitive function, implicating a genotype-related cognitive performance. Furthermore, the novel neuroimaging-transcriptomic analysis further revealed the ALFF alterations being co-localized with the expression profiles of genes enriched in neurodevelopmental and cell migration pathways. ​​This study provided the first evidence linking alteration of brain activity to the molecular underpinnings in patients with KS, supporting the notion of neurodevelopmental genes shaping brain structure and function. This study showed hyperactivity in the left inferior/middle frontal gyrus and insula (Cluster 3), key nodes of the central executive network and salience network. The insula is primarily involved in detecting salient stimuli, regulating attentional switching, and integrating interoceptive and exteroceptive information [40–42]. The increased ALFF observed in these regions may reflect excessive activation related to self-awareness and interoceptive monitoring in patients with KS. Importantly, higher ALFF values in these regions were positively correlated with better global cognition (MoCA) and selective attention (SCWT-B). This pattern may support the “compensatory recruitment” hypothesis. In the context of KS-related neurodevelopmental deficits and sex steroid deprivation, the brain appears to engage in an adaptive upregulation of executive control systems to maintain cognitive function. The bilateral insula and prefrontal cortex are crucial for mobilizing attentional resources and filtering irrelevant stimuli [43, 44]. Therefore, the increased spontaneous activity in Cluster 3 likely represents a successful mobilization of neural resources; patients who can sustain this compensatory hyperactivation are better able to preserve cognitive function despite the underlying pathology. In contrast to the frontal hyperactivity, the left angular gyrus (Cluster 5), a core hub of the default mode network (DMN), exhibited significant hypoactivation. The angular gyrus is indispensable for internally-directed cognitive processes, such as episodic memory retrieval and self-referential processing [45–47]. Paradoxically, within this hypoactive region, higher residual ALFF values were negatively correlated with visual reproduction and digit span (DST-forward) scores. This suggested that the residual or “failure-to-suppress” activity in the angular gyrus may act as neural noise or interference rather than a functional resource. Efficient cognitive processing, particularly for externally directed tasks, typically requires the suppression (deactivation) of the DMN to prevent internal distraction. The negative correlation indicated that patients with KS who could not effectively downregulate this DMN node may suffer from greater interference, leading to poorer memory performance. Furthermore, mediation analysis revealed that the impairment in visual memory significantly mediated the relationship between angular gyrus dysfunction and attentional performance. This supported a potential association with cognitive decline, where foundational dysfunction in the angular gyrus undermined visual memory capacity, and constraining the attentional resources available for working memory tasks. Our exploratory genetic analysis provided initial insight into the biological heterogeneity driving clinical and neuroimaging features observed within the KS cohort. We identified several potential genotype-phenotype correlations: for instance, OTUD4 mutations trended toward an association with better executive function, while FGFR1 mutations were preliminarily correlated to better visual memory. Intriguingly, we identified a potential “genetic dosage” effect, where a higher cumulative mutation burden (Mut_Number) was associated with greater hyperactivity in the compensatory fronto-insular network (Cluster 4). This implied that patients with a more complex genetic background may require more intensive neural compensation to maintain function. Additionally, distinct mutations appeared to map to specific neural profiles; for instance, the PROKR2 mutation was strongly associated with hypoactivation in the angular gyrus (Cluster 5), while ANOS1 mutations were linked to alterations in the cerebellum (Cluster 1). Collectively, despite the statistical limitations of this exploratory approach, these findings provided hypothesis-generating preliminary data that different genetic etiologies likely drive distinct neurodevelopmental trajectories and corresponding brain-cognition profiles, highlighting the necessity of encompassing genetic stratification in future KS neuroimaging research. The successful application of neuroimaging–transcriptomic analysis in this study provided compelling evidence that the spatial patterning of ALFF alterations in patients with KS was significantly constrained by the underlying cortical gene expression. The identification of ANOS1 [5] and WDR11 as top contributors to the PLS1 component offered a validation of our integrative approach. These results demonstrated that brain regions showing abnormal intrinsic activity in patients with KS were also sites of high expression of core disease-related genes. Notably, ANOS1 is a well-established X-linked pathogenic gene for KS, critical for the migration of GnRH neurons and the axonal targeting of olfactory sensory neurons [48, 49]. The strong negative association of ANOS1 expression with ALFF in this study supported the view that disrupted developmental signaling may result in long-lasting suppression of spontaneous neural activity in affected brain regions. WDR11, another gene implicated in KS, is known to regulate ciliogenesis and the development of the olfactory bulb [50, 51], and is essential for GnRH neuronal function via intracellular signaling pathways. The positive association of WDR11 with ALFF, together with its enrichment in intracellular transport and gene regulatory processes, suggested that beyond early developmental defects, subcellular and molecular dysregulation may also contribute to persistent functional abnormalities in patients with KS. Crucially, the genes spatially correlated with ALFF alterations were significantly enriched for pathways governing neurodevelopment, cell migration, and sensory organogenesis. These findings supported the KS as a widespread neurodevelopmental syndrome, where altered intrinsic brain activity may be not a mere epiphenomenon but a direct reflection of genetically-driven brain activity disorder. This multi-level evidence may help to bridge the gap from molecular underpinnings to macroscale brain dysfunction, and offers a framework for understanding the phenotypic heterogeneity of KS, and may ultimately yield biomarkers for early detection and precision-targeted interventions for patients with KS. Several limitations should be considered when interpreting the findings of this study. First, the cross-sectional design limited causal inference, as only associations between brain activity and cognitive function at a single time point could be assessed. However, this study design still provided valuable insights into the neural correlate of KS, and helped to establish a foundation for future longitudinal studies to track developmental trajectories and the impact of hormonal replacement therapy for patients with KS. Second, our KS cohort exhibited a strong male predominance. While this may have reflected this disorder with male dominance, the limited number of female patients in this cohort precluded meaningful subgroup analysis of sex-related differences in neurocognitive manifestations. Future multicenter studies should aim to enroll more female patients to explore potential sex dimorphism in KS-related brain alterations. Third, the KS cohort for this study was genetically heterogeneous, which may have reflected the real-world clinical presentation of the disorder. Our exploratory genotype–phenotype analysis provided preliminary data to disentangle the genetic heterogeneity. However, this analysis was constrained by the relatively small sample size that was underpowered, as compared to the greater number of genetic and phenotypic variables tested. Furthermore, because we did not apply multiple comparison corrections to avoid missing potential early signals, the genetic findings identified in this study may carry a risk of false positives and therefore must be interpreted with strict caution. A larger, genetically stratified cohort should further clarify how specific mutations shape brain function and cognition. Fourth, the neuroimaging–transcriptomics analysis relied on the Allen Human Brain Atlas, which was derived from a small number of healthy adult donors and may not be the most appropriate atlas for this study of patients with KS. However, our study also identified the known KS-related genes (e.g., ANOS1, WDR11), which strongly validated the biological relevance of our study findings. Despite these limitations, our study had notable strengths. It represented one of the largest neuroimaging cohorts of KS to date with 100 patients with KS enrolled into the study. In addition, this study adopted a multimodal approach integrating functional neuroimaging, cognitive testing, genetic assessment, and transcriptomics to assess the interplay among neuroimaging, genetics and function. This study contributed novel evidence to support KS be defined as a neurodevelopmental syndrome with molecular underpinning. In summary, this study demonstrated that KS was associated with a profound dysregulation of intrinsic brain activity, characterized by a trade-off between compensatory executive hyperactivation and dysfunctional DMN suppression. These functional alterations were directly linked to cognitive impairments and were modulated by the underlying genetic burden. These findings implicated ALFF alterations as promising neuroimaging biomarkers for cognitive risk stratification in patients with KS. Supplementary Information Supplementary Material 1 Acknowledgements We sincerely thank the staff members from the Department of Radiology at Xiangya Hospital, for their invaluable support and collaboration in patient recruitment, clinical assessment, and imaging data acquisition. We also wish to thank all the patients and healthy volunteers whose participation made this study possible. Abbreviations Amplitude of Low-Frequency Fluctuation Blood-Oxygen-Level-Dependent Digit Span Tests Echo Time False Discovery Rate Gonadotropin-Releasing Hormone Gene Ontology Gaussian random field Kallmann Syndrome Kyoto Encyclopedia of Genes and Genomes Montreal Cognitive Assessment Resting-state Functional Magnetic Resonance Imaging Repetition Time Stroop Color and Word Test Trail Making Test Visual Reproduction Authors’ contributions Yan Fu and Xueying Wang: Writing-original draft, Writing-review & editing, Data collection. Jinfan Zhang, Feifei Wu, Yafang He, Han Wu, and Yuanhao Li: Writing-review, Data collection. Meichao Men: Study design, Writing-review & editing, Conceptualization. Yuanchao Zhang: Study design, Writing-review & editing, methodology. Alessandro Grecucci: Writing-review & editing. Xiaoping Yi: Study design, Writing-review & editing, Conceptualization. Jiada Li and Wenzhen Zhu and Bihong T. Chen: Writing-review & editing, Conceptualization. Funding This research was funded in part by the Scientific Research Program of FuRong Laboratory (No.2024PT5109), the Joint Research Fund of Chongqing University Three Gorges Hospital-Peking University- Hunan Dajiaweikang Pharmaceutical Industry Co., Ltd. (IIT2025-00111). Data availability All relevant data are reported in the article. Additional details can be provided by the corresponding author upon reasonable request. Declarations Ethical approval and Consent to participate The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013) and approved by the Institutional Review Board of Xiangya Hospital (No. 2022101052). All participants provided written informed consent. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Meichao Men, Email: menmeichao@csu.edu.cn. Yuanchao Zhang, Email: yuanchao.zhang8@gmail.com. Xiaoping Yi, Email: yixiaoping@cqu.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement All relevant data are reported in the article. Additional details can be provided by the corresponding author upon reasonable request.