Brain structural-functional connectivity coupling and glucose metabolism in adolescents with bipolar disorder and major depressive disorder

Abstract

Aim

To explore potential neuroimaging features derived from [¹⁸F]FDG PET/MRI imaging that may help characterize differences between adolescent with bipolar disorder (BD) and those with major depressive disorder (MDD).

Materials and methods

Patients were grouped based on the Hamilton Depression Scale (HAMD) and Hypomania Check List (HCL), and the severity of suicidal ideation and behavior was assessed. 14 BD and 14 MDD adolescent patients were enrolled. Simultaneous PET and MRI data (including 3D T1-weighted imaging, Diffusion Tensor Imaging (DTI), and resting-state functional MRI (fMRI)) were acquired. Functional connectivity (FC) and structural connectivity (SC) were derived from fMRI and DTI respectively. A multiple linear regression model was employed to predict FC from SC. SC-FC coupling was defined as Pearson correlation coefficient between predicted FC and original FC. Using cerebellar gray matter uptake as reference region, standardized uptake value ratio (SUVr) of each brain region was calculated as metabolic parameter.

Results

Compared with the MDD group, the BD group showed significantly decreased SC-FC coupling and glucose metabolism in the thalamus. ‌In the BD group, SC-FC coupling in the paracentral lobule showed a significant positive correlation with metabolism (r = 0.70). In the MDD group, SC-FC coupling in cingulate gyrus showed a significant negative correlation with metabolism (r=-0.73). ‌In the BD group, SC-FC coupling in superior parietal lobule, postcentral gyrus, and precentral gyrus correlated significantly with HAMD, HCL, and suicide scores, respectively. In the MDD group, SC-FC coupling in precentral gyrus, orbital gyrus, and hippocampus showed the highest correlations with these clinical indicators. The middle temporal gyrus and parahippocampal gyrus were specifically associated with suicide scores in the BD group. ‌Notable intergroup differences between BD and MDD were thalamus (AUC = 0.86) and inferior temporal gyrus (AUC = 0.76), with SC-FC coupling features in these regions showing more prominent differences than metabolic features.

Conclusion

This study identifies the thalamus as a key region in adolescent BD. Differences in SC-FC coupling in thalamus and inferior temporal gyrus reflect notable exploratory features between adolescent BD and MDD patients, suggesting that SC-FC coupling parameters may represent potential neuroimaging signatures for differential diagnosis. The differences in correlations among SC-FC coupling, metabolism, and clinical parameters between the groups indicate distinct pathogenic mechanisms for BD and MDD.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s00259-026-08023-w.

Introduction

Suicide is a major public health problem and the second leading cause of death among individuals aged 10–24 years globally [1]. BD is one of the core risk factors for adolescent suicide, with a suicide risk ten times higher than that of the general population and an incidence of up to 1.8% in this age group [2]. However, adolescent BD is prone to being misdiagnosed as MDD or manifestations of the “rebellious phase”, primarily due to the considerable overlap of symptoms among these conditions and underrecognized cognitive differences. The manic/hypomanic episodes of BD are often misidentified as normal adolescent mood swings or attention-deficit/hyperactivity disorder (ADHD). Some studies have confirmed that adolescent BD is frequently misclassified: up to 33% of adolescent BD patients are initially misdiagnosed with MDD, while manic/hypomanic symptoms are often mistaken for ADHD or normal mood lability [3]. This diagnostic challenge stems not only from clinical oversight but also from the profound cognitive impairments inherent to BD—including deficits in executive function, emotion regulation, and insight—which blur clinical boundaries with MDD, ADHD, and typical adolescent development [4, 5]. The clinical manifestations of depressive episodes in BD share considerable overlap with those of MDD such as markedly depressed mood, decreased energy, and social withdrawal, and others [6]. Additionally, patients tend to seek medical help during depressive rather than manic episodes [7], which may lead to underreporting of hypomanic symptoms and hindering accurate diagnosis [4]. Furthermore, the subjective nature of psychiatric evaluations of patient self-reported symptoms also contributes to misdiagnosis. Once adolescent BD is misdiagnosed as MDD, early antidepressant drug treatment is likely to induce mixed episodes [8]. Frequent mood swings may further disrupt patients’ cognitive function and elevate suicide risk [8, 9]. Therefore, the early identification of adolescent BD constitutes an important clinical issue. Current diagnosis relies on retrospective medical histories, resulting in diagnostic delays of 5–10 years in adolescent patients [10, 11]. However, no effective physiological indicators for the early diagnosis of BD are currently available.

Neuroimaging technology has provided unprecedented opportunities for elucidating the anatomical correlates of MDD. Pathological alterations of metabolism in brain structures implicated in emotional/cognitive function in depression have been identified as significant contributors to the pathological mechanisms of MDD [12], particularly supporting a role for dysfunction within the prefrontal cortex and striatal systems that normally modulate limbic and brainstem structures involved in mediating emotional behavior in the pathogenesis of depressive symptoms. An earlier meta-analysis [13] showed that the activation of the prefrontal cortex was attenuated in the manic state of BD, while the activation of the limbic system remained consistently enhanced across all the emotional states, and the part responsible for impulse inhibition is damaged, resulting in emotional dysregulation and weakened behavioral inhibition ability. In patients with BD-II with suicidal behavior, the functional activities of the anterior cingulate gyrus and middle frontal gyrus, and the FC between them, are significantly reduced [14]. Abnormalities in these neural circuits make it difficult for patients to effectively regulate their emotions and control impulsive behaviors when faced with various stimuli. Current studies have also shown that brain glucose metabolism may serve as a valuable physiological indicator for assessing neuronal activity and holds great significance in elucidating neurophysiological mechanisms underlying the increased risk of depression in young adults. In particular, the posterior cingulate cortex (PCC) exhibits high connectivity and metabolic activation level [15], in which metabolic activity has been indicated as a significant neuroimaging biomarker of depression symptoms and antidepressant treatment outcome [16].

Previous research has predominantly utilized healthy populations as control groups, and there are relatively few reports on neuroimaging analyses of regional brain differences between depressive episodes in BD and MDD – two affective disorders with highly similar early clinical symptoms that are easily confused. Furthermore, most prior studies have primarily focused on single-parameter analyses, such as SC, FC, or glucose metabolism. Consequently, the reported results exhibit considerable variability and may lack robustness [13]. Therefore, our study addresses the uniqueness of adolescent first-episode BD and MDD samples and is based on the following hypotheses. First, the clinical symptom profile of BD in adolescents differs from that of MDD, with more severe depressive and anxiety symptoms. Second, structural-functional connectivity and glucose metabolism in core brain regions (e.g., prefrontal cortex, cingulate cortex, thalamus, hippocampus) differ between adolescents with BD and those with MDD. By integrating multimodal data, including clinical symptoms, fMRI-based structural-functional connectivity, and PET-based glucose metabolism, we aim to identify differences in neuroimaging characteristics between depressive episodes in BD and MDD in adolescents, and expect that this research will address the limitations of previous studies and develop early differentiation methods to facilitate timely intervention.

Methods

Participants

Patients were recruited from the Psychiatric inward of the Shanghai Tenth People’s Hospital, China. The first participant was enrolled on June 15, 2024, and the last participant was enrolled on February 24, 2025. All enrolled participants are scheduled for followed-up assessments at 1 year, 2 years, or beyond. The diagnoses of BD and MDD were determined based on the diagnostic criteria of Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5). For patients with BD, it should be emphasized that participants recruited into the BD group were diagnosed with BD-II at the time of enrollment, when PET/MR scanning was concurrently conducted. Specifically, eligible subjects had experienced at least one hypomanic episode and one major depressive episode, with no prior manic episodes, and were in a current depressive episode upon enrollment. Their clinical symptoms were evaluated using the 17-item Hamilton Depression Scale (HAMD-17) and the Hypomania Check List (HCL-32), while patients with MDD were assessed with the HAMD-17. A HAMD score ≥ 14 was indicative of significant depressive symptoms, and a HCL score ≥ 14 was suggestive of significant hypomanic symptoms. Participants who met both score criteria (HAMD ≥ 14 and HCL score ≥ 14) were assigned to the BD group, while those only satisfied the HAMD score criterion (HAMD ≥ 14) were allocated to the MDD group. Meanwhile, the severity of suicidal ideation and behavior of all patients were also assessed. The raw score (ranging from 0 to 4 points) of Item 3 (Suicidal Ideation) on the HAMD-17 was converted into a dichotomous variable: scores of 0–3 were defined as “0” (no suicidal behavior), and a score of 4 was defined as “1” (presence of suicidal behavior). The inclusion and exclusion criteria for all participants were shown in Supplementary Fig. 1. Prior to enrollment, all provisional cross-sectional baseline diagnoses were independently reviewed by an associate chief physician to rule out comorbid psychiatric or neurological disorders. Subsequent psychological assessments were conducted by trained and experienced psychometricians.

Image acquisition and image processing

[¹⁸F]FDG PET and MRI were simultaneously acquired by using a hybrid 3.0-Tesla TOF-PET/MRI scanner (uPMR 790, United Imaging, Shanghai, China) with a 24-channel head and neck coil. Detailed information on PET scanning protocols, subject preparation, MR acquisition sequences (including high-resolution 3D T1-weighted imaging, DTI, and resting-state fMRI), as well as image processing pipelines (T1-weighted image processing, resting-state fMRI processing and FC construction, DTI processing and SC construction, SC-FC coupling calculation, and PET data processing) is available in the Supplementary Materials and Methods.

Statistical analysis

All statistical analyses were performed using Matlab R2022a. The differences of SC-FC coupling and metabolism in brain regions between adolescent patients with BD and those with MDD were examined by two-tailed two-sample t-tests, gender and age were included as covariates. To assess the association between SC-FC coupling and metabolism, we calculated partial Pearson’s correlation coefficient with age and gender as covariates. For the correlation of SC-FC coupling with clinical metrics, partial Spearman’s correlation coefficient was calculated, with adjustment for age and gender. We also used the logistic regression model to evaluate the ability of SC-FC coupling and metabolic parameters to discriminate between adolescent patients with BD and those with MDD. About the general demographic data and symptom scale scores, according to Shapiro-wilk normal test, the scores of ages, years of education, HAMA, HAMD, HCL and other scales are normally distributed. The only child or not and gender are counted as data. Independent sample t test and fisher exact test were used for statistical analysis respectively.

Results

Sociodemographic and clinical characteristics of 14 participants with BD and 14 with MDD are summarized in Table 1. The results indicated no statistically significant differences between the two groups in terms of gender, age, education level and whether they were children without siblings (both P > 0.05). In contrast, significant intergroup differences were observed in the HAMD and HCL scores, with all scores in the BD group being significantly higher than those in the MDD group (all P < 0.01). These findings suggested that the severity of depression was significantly greater in adolescent patients with BD than in the MDD group.

Table 1.

General demographic data and symptom scale scores (øverline x± s; Number )

	BD group	MDD group	t/χ2	p
Age (years)	14.28 ± 1.68	15.00 ± 2.07	-1.00	0.32a
Educational level (years)	8.85 ± 1.61	9.50 ± 2.37	-0.84	0.41a
Gender (Male/female)	0/14	4/10	--	0.09b
The Only child (Yes/No)	12/2	8/6	--	0.21b
HAMD	19.64 ± 3.45	10.36 ± 6.62	4.65	0.00a
HCL	19.07 ± 3.36	11.57 ± 7.33	3.48	0.00a

^ais independent sample t-test, t value; ^bis Fisher’s exact test

SC-FC coupling differences between BD and MDD

Compared with the MDD group, the BD group exhibited decreased SC-FC coupling in 6 brain subregions and increased SC-FC coupling in 2 subregions (Fig. 1A, Supplementary Table 1). The middle temporal gyrus (p = 0.005, T=-3.06), inferior temporal gyrus (p = 0.003, T=-3.26), and thalamus (p = 0.020, T=-2.50) showed significantly decreased SC-FC coupling in the BD group relative to the MDD group, while the cingulate gyrus (p = 0.014, T = 2.64) showed significantly increased SC-FC coupling in the BD group compared with the MDD group (Fig. 1B). However, all these results did not pass the false discovery rate (FDR) correction.

Fig. 1.

Structural and Functional connectivity (SC-FC) coupling differences between BD and MDD groups. Compared with MDD, the BD group exhibited more significantly reduced SC-FC coupling in the MTG, ITG and thalamus, as well as more significantly increased SC-FC coupling in the CG. (MTG, Middle Temporal Gyrus; ITG, Inferior Temporal Gyrus; CG, Cingulate Gyrus; Tha, Thalamus)

Metabolism differences between BD and MDD

Brain-region-level metabolic differences between the BD and MDD groups are shown in Fig. 2 and Supplementary Table 2. The BD group showed consistent metabolic decrease in statistically different brain regions. The striatum (p < 0.05), angular gyrus (p = 0.010, T=-2.80), and thalamus (p = 0.006, T=-2.97) displayed the most prominent metabolic reductions (Fig. 2B).

Fig. 2.

[¹⁸F]FDG metabolism differences between BD and MDD. (A, B) Compared with MDD, the BD group exhibited more significantly reduced metabolism in the Striatum, Angular Gyrus and Thalamus. (IPL, Angular Gyrus; Str, Striatum; Tha, Thalamus). (C) presents the group-averaged FDG images of both BD and MDD groups

Association analysis between SC-FC coupling and metabolism

Using partial Pearson’s correlation analysis to evaluate the association between SC-FC coupling and metabolism, we found that in the BD group (Fig. 3, Supplementary Table 3), the maximum positive correlation was observed in the paracentral lobule (p = 0.011, r = 0.70), and no significant inverse association was found. In the MDD group (Fig. 3, Supplementary Table 4), the maximum positive correlation was found in the thalamus (p = 0.023, r = 0.67), and the maximum inverse association was identified in the cingulate gyrus (p = 0.010, r=-0.73). However, none of the regional correlations remained significant after corrections for multiple comparison.

Fig. 3.

Brain regions showing significant associations between SC-FC coupling and metabolism in the BD and MDD groups

Correlation between SC-FC coupling and clinical metrics

Partial Spearman’s correlation analysis was performed to examine the correlation between SC-FC coupling and clinical metrics, including the HAMD, HCL, and suicide scores. The results are shown in Fig. 4 and Supplementary Tables 5–10. In the BD group, the maximum correlation with HAMD was in the superior parietal lobule (p = 0.010, r=-0.68), and the maximum correlation with HCL was in the postcentral gyrus (p = 0.001, r=-0.80), and the maximum correlation with suicide scores was in the precentral gyrus (p = 0.000, r=-0.96). In the MDD group, the maximum correlation with HAMD was in the precentral gyrus (p = 0.003, r=-0.86), and the maximum correlation with HCL was in the orbital gyrus (p = 0.005, r = 0.83), and the maximum correlation with suicide scores was in the hippocampus (p = 0.000, r = 0.96). Among the brain regions with intergroup differences in SC-FC coupling, only the middle temporal gyrus (p = 0.015, r=-0.65) and the parahippocampal gyrus (p = 0.012, r=-0.67) in the BD group showed a significant correlation with suicide scores.

Fig. 4.

Brain regions showing significant association between SC-FC coupling and clinical metrics in the BD and MDD groups

Receiver operating characteristics analysis

In the receiver operating characteristics (ROC) analysis, the characteristic brain regions were derived from the differential brain regions between the groups in SC-FC coupling and metabolism, and we averaged the sub-regions that were attributed to the same brain region as a characteristic brain region. The optimal cut-offs were chosen to maximize Youden’s index. The inferior temporal gyrus (AUC = 0.863, cut-off = 0.195 ± 0.033, sensitivity = 0.775 ± 0.208, specificity = 0.960 ± 0.104, accuracy = 0.868 ± 0.101) and the thalamus (AUC = 0.764, cut-off = 0.25 ± 0.03, sensitivity = 0.700 ± 0.235, specificity = 0.970 ± 0.081, accuracy = 0.835 ± 0.110) in SC-FC coupling demonstrated relatively high discriminatory ability for distinguishing between BD and MDD. Metabolic profiles exhibited substantially lower discriminatory power between the two groups compared with SC-FC coupling features (Fig. 5).

Fig. 5.

Receiver operating characteristics (ROC) analysis

Subsequently, we focused on the inferior temporal gyrus (ITG) and the thalamus to analyze the specific differences in FC and SC between these two regions and core emotion-regulatory brain regions (prefrontal cortex, hippocampus, amygdala and cingulate gyrus) between the BD and MDD groups. Our findings revealed no significant intergroup differences in FC between the ITG and the aforementioned brain regions. In contrast, significant intergroup differences were detected in SC between the ITG and the prefrontal cortex (Precentral Gyrus PrG_L_6_1, p = 0.001), with a marked enhancement of ITG-prefrontal cortical SC observed in BD patients relative to MDD patients. Intergroup differences were observed in the FC between the thalamus and the prefrontal cortex as well as the hippocampus (specifically: Middle Frontal Gyrus MFG_L_7_1, p = 0.003, MFG_L_7_3, p = 0.004, Orbital Gyrus OrG_L_6_3, p = 0.011, Hippocampus Hipp_L_2_2, p = 0.036, Hipp_R_2_2, p = 0.048). Specifically, the thalamus-prefrontal cortex FC was significantly higher in the BD group than in the MDD group, whereas the thalamus-hippocampus FC was lower in the BD group relative to the MDD group. In addition, intergroup differences were identified in the SC between the thalamus and the prefrontal cortex (specifically: Middle Frontal Gyrus MFG_R_7_3, p = 0.035, Orbital Gyrus OrG_R_6_2, p = 0.039, Precentral Gyrus PrG_R_6_5, p = 0.043). Concretely, the thalamus-MFG_R SC was higher in the BD group than in the MDD group, while the thalamus-OrG_R SC and thalamus-PrG_R were lower in the BD group compared with the MDD group.

Discussion

Regarding the clinical HAMD and HCL scores, our study results showed that all scores in the adolescent BD group were higher than those in the MDD group. This indicates that adolescent BD patients in depressive episodes experience more severe depressive symptoms compared with MDD patients-a pattern consistent with the hallmark features of adolescent BD. These features include acute dysregulation of emotion, behavior, and cognitive function, often accompanied by high-risk impulsivity and suicidality [11]. Consequently, early identification of BD is critically important. Furthermore, regarding neuroimaging findings, our results revealed significantly reduced SC-FC coupling in the middle temporal gyrus (MTG), inferior temporal gyrus (ITG), and thalamus in BD patients compared with MDD patients. The thalamus, a critical hub for sensory integration and consciousness regulation, plays a pivotal role in BD pathophysiology.

Previous structural imaging studies of BD identified decreased volumes particularly in the thalamic and hippocampal regions [17]. Of these changes, volumetric alterations in the thalamus seemed to be specific to BD [18]. Our findings align with these structural observations: reduced thalamic SC-FC coupling in adolescent BD may stem from thalamic volume loss, which impairs affective filtering and increases sensitivity to external stimuli, thereby exacerbating emotional lability. Conversely, thalamic dysfunction may further accelerate volumetric decline. Baxter LR Jr [19] and Lei Zhang [20] reported significantly lower thalamic glucose metabolism in BD patients, especially during depressive episodes, while Tu PC [21] noted thalamocortical dysconnectivity in both BD and MDD, with lower thalamocortical FC in BD patients than in MDD patients. Through direct comparative analysis between adolescents with BD and MDD, our study further demonstrated that, in addition to abnormal FC, adolescent patients with BD exhibited combined abnormalities in thalamic–prefrontal SC, SC-FC coupling, and glucose metabolism. These findings collectively support a critical role of the thalamus in the neuropathology of BD, with all the aforementioned measures being lower in patients with BD than in those with MDD. In addition, MDD patients showed a positive association between thalamic SC-FC coupling and glucose metabolism (p = 0.023, r = 0.67), suggesting aligned structural-functional and metabolic states. Most previous relevant studies used healthy volunteers as controls: Saxena S [22] reported significantly elevated bilateral thalamic glucose metabolism in MDD patients. Xu M [23] analyzed SC-FC coupling in 168 adolescents with MDD and 101 healthy controls, demonstrating enhanced left-thalamic coupling in MDD patients without self-injurious behaviors. Zhang Y [24] identified strengthened structural connectivity in the left thalamus but weakened connectivity in the right thalamus. In contrast, through direct comparison between patients with BD and those with MDD, combined with previous findings, our study preliminarily observed key differences in thalamic neuroimaging characteristics between the two disorders. Distinct from the consistently reduced combined thalamic abnormalities in BD, patients with MDD exhibited more pronounced heterogeneity in thalamic structural-functional connectivity and glucose metabolism (i.e., either increased or decreased). This finding highlights a notable distinction between our study and prior research. Based on existing evidence, we speculate that this thalamic heterogeneity in MDD may be associated with impaired processing of negative stimuli [25].

Given that significant differences exist in thalamic structural-functional connectivity and glucose metabolism between adolescent patients with BD and those with MDD, does this disparity hold potential value for revealing the distinctive neurobiological underpinnings of the two disorders? The current results of receiver operating characteristic (ROC) analysis showed that thalamic SC-FC coupling exhibited a certain classification efficacy for distinguishing BD in depressive episodes from MDD (AUC = 0.76). Although the current results in Sect. Metabolism differences between BD and MDD showed that thalamic glucose metabolism was lower in the BD group than in the MDD group, the ROC analysis revealed that the difference in thalamic glucose metabolism between the two groups has limited value for differential diagnosis. Furthermore, significant intergroup differences between the adolescent patients with BD and those with MDD emerged in thalamic-prefrontal functional connectivity (FC) (middle frontal gyrus [MFG], orbital gyrus [OrG]), structural connectivity (SC) (MFG, OrG, precentral gyrus [PrG]), and thalamic-hippocampal FC. As a hub for dynamic network switching, the thalamus forms dense bidirectional fiber connections with the prefrontal cortex through specialized nuclei, creating a highly integrated core network for advanced cognition, emotion regulation, and behavioral control via thalamic relay, prefrontal integration, and behavioral output. Concurrently, the thalamic-hippocampal circuit underpins core memory-emotional processing. This aligns with the findings of Lu F [26], who reported predominant thalamic-limbic and striatal circuit abnormalities in MDD, contrasting with broader thalamic dysregulation in BD. Collectively, thalamic alterations represent a critical neural mechanism in BD, functionally validating structural findings reported by Ding JB [18]. Thus, significantly reduced thalamic SC-FC coupling may represent a distinct characteristic for BD-MDD differentiation, with an accuracy of 83.5%.

Simultaneously, our results demonstrated that SC-FC coupling in the ITG of adolescent patients also exhibited certain discriminatory power in distinguishing depressive episodes of BD from MDD (AUC = 0.86). The ITG serves as a core region for high-level visual processing and collaborates with limbic structures (amygdala, hippocampus) in affective memory integration. Previous studies have indicated that BD patients exhibit ITG volume expansion during early manic phases, but progressive gray matter thinning in chronic stages [17], compromised white matter tract integrity, and structural decoupling from limbic networks. These alterations contribute to manic hyperexcitability, enhanced negative memory bias during depression, and functional disconnection from prefrontal circuits [27]. Consequently, amplified processing of negative valence visual stimuli (e.g., sad facial expressions) impairs cognitive control over emotional impulses. In contrast, MDD patients show persistent ITG volume reduction, though reports of specific white matter disruptions are limited [28], with greater emphasis on default-mode network dysconnectivity [24]. Notably, our results revealed no intergroup differences in ITG functional connectivity (FC) with the prefrontal cortex, hippocampus, amygdala, or cingulate gyrus between adolescent patients with BD and those with MDD. However, BD patients exhibited significantly higher ITG-prefrontal structural connectivity (SC), yet lower ITG SC-FC coupling compared with MDD patients. This dissociation suggests impaired effective functional integration—specifically, disrupted visuospatial processing and semantic memory consolidation—which may degrade top-down, goal-directed attentional processes [29, 30]. Such deficits appear more pronounced in adolescent BD, likely exacerbating cognitive dysfunction and emotional dysregulation. In our cohort, the classification accuracy of ITG reached 86.8%, aligning with the finding of Tan [31], who reported disrupted ITG-dorsal attention network coherence in BD (accuracy > 60%). These collective findings further support the ITG’s pivotal role in BD pathogenesis and its potential value in characterizing differences between adolescent depressive episodes in BD patients and MDD patients.

At rest, SC-FC coupling reflects positive alignment between structural connectivity strength and functional coherence [32], while glucose metabolism directly indexes neuronal activity. In the MDD group, the thalamus showed the strongest positive SC-FC–metabolism correlation, while the cingulate gyrus (CG) exhibited the most significant negative association. These observations align with previous reports of greater thalamic connectivity variability in MDD (including both hypo- and hyperconnectivity) [33]. A meta-analysis [34] of cerebral glucose metabolism in MDD patients revealed decreased glucose metabolism in the CG but increased glucose metabolism in the thalamus. This pattern is associated with compensatory hyperactivity in the thalamus. Hosokawa T [35] differentiated metabolic patterns: BD showed hypometabolism in the right anterior CG (a region implicated in emotional processing), whereas MDD involved the left posterior CG (a key hub of the default-mode network associated with cognitive impairment). Our results demonstrated significantly higher SC-FC coupling in the left anterior cingulate gyrus (LAC) in the BD group compared with the MDD group, despite that no significant intergroup differences in glucose metabolism were observed. In summary, as a core limbic region, the cingulate gyrus exhibits distinct profiles in structural-functional connectivity and glucose metabolism between BD and MDD patients. Previous research indicates more complex cingulate pathophysiology in BD, necessitating further investigation into its differential involvement across these disorders. Conversely, adolescent BD patients demonstrated the strongest positive correlation between SC-FC coupling and metabolism in the paracentral lobule (PCL). Clinically, we observed more severe motor-executive impairments in BD than in MDD. Given the PCL’s role as a somatomotor network hub governing lower-limb sensation and movement, we postulate that concurrent structural-functional abnormalities in this region underlie BD-specific motor deficits.

Although MDD and depressive episodes in BD share core depressive symptoms, they exhibit distinct clinical manifestations and disease course characteristics. When examining correlations between brain SC-FC coupling and depressive symptoms (assessed by HAMD scores), significant associations emerged in distinct regions: the superior parietal lobule (SPL) in BD patients, and the PrG in MDD patients. The SPL supports high-level cognitive functions encompassing spatial awareness, attentional allocation, working memory, and multisensory integration [36]. Its correlation with depressive symptoms in BD may reflect cognitive abnormalities during depressive episodes—specifically, impaired sustained attention, task-focused persistence, and diminished working memory capacity [37]. These deficits align with the SPL’s role in attentional and mnemonic processing, suggesting that structural-functional decoupling in this region contributes to BD-specific cognitive disturbances. Previous resting-state fMRI studies [38] have demonstrated that in BD patients, reduced gray matter volume in the SPL, disrupted white matter integrity, and phasic fluctuations in dynamic FC (e.g., connectivity enhancement during manic phases and attenuation during depressive episodes) can distinguish BD from MDD and may consequently serve as a diagnostic biomarker for BD. Conversely, PrG dysregulation in MDD patients disrupts motor planning and execution [39], thereby manifesting as psychomotor retardation (e.g., slowed movement, reduced speech)—a hallmark MDD symptom.

Correlation analyses with the HCL scores revealed distinct neural substrates in BD versus MDD patients. In BD patients, altered SC-FC coupling in the postcentral gyrus (PoG) significantly correlated with hypomanic symptoms. As a primary somatosensory region, the PoG mediates sensory information processing [40]. We propose that during hypomania, aberrant somatic states (e.g., altered muscle tone, tachycardia, hypertension) generate excessive sensory input, potentially driving SC-FC reorganization in this region [41, 42]. Conversely, MDD patients exhibited orbital gyrus (OrG) abnormalities correlating with HCL scores. The OrG forms extensive connections with limbic structures (amygdala, hippocampus) and prefrontal areas, constituting an emotion-cognition regulatory network [43] responsible for affect modulation, behavioral inhibition, and social cognition. Although MDD patients lack true hypomanic episodes, transient subclinical activation may occur during symptom fluctuations, manifesting as subtle increases in speech and behavioral activity. OrG dysregulation in MDD may disrupt this network’s equilibrium, generating behavioral deviations with weak phenomenological parallels to hypomania.

Our study aims to explore neuroimaging differences between depressive episodes of BD patients and MDD patients, with the goal of enabling early identification and preventing the progression of suicidal ideation into concrete actions at the critical stage. Correlation analyses targeting suicidality revealed divergent hippocampal involvement in BD versus MDD. In MDD patients, abnormal SC-FC coupling in the hippocampus significantly correlated with suicidal behaviors. As a key region for memory consolidation, emotion regulation, and stress response [44], the hippocampus undergoes volume loss and neuronal dysfunction under chronic depression, which impairs its core operations. We propose three potential mechanisms. First, hippocampal dysfunction disrupts the retrieval and integration of positive memory, trapping patients in negative recall cycles that amplify hopelessness [45–47]. Second, aberrant hippocampal-limbic connectivity dysregulates affective processing, hindering adaptive responses to negative emotions [47, 48]. Third, maladaptive stress-signal processing [49] heightens vulnerability to suicidal ideation during life stressors. Conversely, BD patients exhibited suicide-related SC-FC abnormalities in the precentral gyrus (PrG)—a primary motor cortex governing action execution [50]. We hypothesize that structural-functional hyperactivity in this region during mania drives impulsive suicidal acts, while disrupted prefrontal-PrG circuitry impairs inhibitory control over dangerous behaviors, facilitating rapid transition from suicidal ideation to concrete action [51]. Collectively, these distinct neural substrates highlight opportunities for developing exploratory models to precisely stratify suicide risk across mood disorders.

Our study revealed significant correlations between SC-FC coupling and clinical symptoms across multiple brain regions in adolescent BD patients and MDD patients. Crucially, significant correlations between SC-FC coupling and suicide scores were observed exclusively in the BD group, specifically in the middle temporal gyrus (MTG) and parahippocampal gyrus (PHG). As demonstrated in Supplementary Table 1, BD patients exhibited markedly reduced MTG SC-FC coupling relative to MDD counterparts. Given the MTG’s established role in cognition, we posit that MTG dysfunction severely disrupts processing of external stimuli and emotional perception in BD, thereby elevating suicide risk. A meta-analysis by Chen CF [52] on whole-brain fMRI abnormalities in suicidal individuals identified hyperactivation in the left MTG among suicide attempters. In an emotion-matching task fMRI study [53], BD patients exhibited hypoactivation in MTG compared with healthy controls. Additionally, studies have demonstrated significantly reduced fALFF [54] and FC [55] strength in the right MTG of BD patients. Similarly, the aforementioned Tan’s study [31] revealed significantly decreased FC in the right MTG within the ventral attention network of BD patients. Our findings regarding MTG SC-FC coupling align with this prior evidence. Notably, Supplementary Table 1 also indicated slightly elevated PHG SC-FC coupling in BD patients versus MDD patients. The PHG processes emotion-laden memories via hippocampal circuitry and modulates hypothalamic-pituitary-adrenal axis activity, influencing stress response and emotional stability. In the fMRI study examining SC-FC coupling alterations in adolescent MDD, Xu M [23] analyzed 168 MDD patients and 101 healthy controls, revealing specifically reduced PHG coupling in those with prior suicide attempts. In BD patients, evidence indicates [56] that subcortical structures and limbic systems (e.g., caudate nucleus, amygdala, PHG) exhibit hyperactivation during emotion-related tasks. Notably, the PHG emerges as one of the core dysfunctional brain regions in BD patients with suicidal tendencies [57]. Conversely, during memory retrieval tasks, BD patients demonstrated hypoactivation in the hippocampus and PHG [58]—a pattern that mirrors our finding. Thus, while PHG function in depressive episodes in BD may resemble MDD-like suppression, manic phases drive pathological hyperactivation, precipitating excitation-driven impulsivity—including suicidal behaviors—that indicate the different mechanisms of different suicidal choices in BD.

Our study employed fMRI and PET/MR techniques to assess differences in SC-FC coupling and glucose metabolism between MDD adolescent patients and depressive episodes in BD adolescent patients; our sample was well-matched, yet several limitations warrant acknowledgment. First, the diagnosis of adolescent BD-II is temporally uncertain and unstable over the course of the illness [59, 60]. In the present study, all adolescent participants diagnosed with BD-II received a preliminary cross-sectional baseline diagnosis at enrollment in accordance with DSM-5 criteria. This was a single time-point diagnosis at study entry, rather than a definitive lifelong diagnosis confirmed by long-term follow-up. PET/MR data acquisition and analysis were also completed at the same time point. Given the poor diagnostic stability of adolescent BD-II without long-term longitudinal monitoring, we acknowledge that the interpretation of our findings requires caution. Nevertheless, our results reflect brain function in adolescents currently diagnosed with BD-II at the baseline assessment stage. In addition, although the article title refers generally to BD, our study cohort included only BD-II patients, which may restrict the generalizability of our cross-sectional findings. This is attributable to the scarcity and extreme difficulty in recruiting adolescents across the full bipolar spectrum who meet the enrollment criteria. We are currently conducting ongoing follow-up to dynamically track diagnostic changes in participants. In future research, subjects with ambiguous or revised diagnoses during follow-up will be excluded to establish a standardized long-term follow-up cohort. Meanwhile, we will continue to expand our BD sample bank and strive to include more subtypes across the bipolar spectrum, to further validate and generalize the findings of this study. Second, sample-related constraints: the modest sample size (n = 28) may compromise statistical robustness, potentially limiting comprehensive detection of SC-FC and metabolic differences between groups, as well as the correlation between regional neural variations and clinical parameters. Gender imbalance (with a female predominance) may introduce confounding effects. Additionally, subjective factors in scale-based clinical assessments cannot be ruled out. Third, methodologically, atlas-based brain parcellation during SC-FC coupling construction may obscure individual structural variations; future studies should validate findings at the individual level. The causal interplay among SC, FC, and coupling metrics remains ambiguous. While glucose metabolism provides complementary functional insights, it is insufficient to delineate tripartite interactions between SC, FC, and coupling dynamics. Furthermore, emerging evidence suggests that the cerebellum may be involved in emotional processing and regulation [61]. In the present study, we used the cerebellum as the reference region and performed semi-quantitative analysis of [¹⁸F]FDG uptake, owing to the practical constraints of dynamic PET scanning and arterial blood sampling. These methodological choices should also be acknowledged as potential limitations of our study. Future investigations should employ anatomically refined subdivisions in key brain regions (e.g., thalamus, temporal subregions) for granular coupling analyses.

Conclusion

In summary, adolescents with MDD and adolescents experiencing depressive episodes of BD exhibit distinct regional differences in SC-FC coupling and glucose metabolism. Compared with MDD patients, adolescent BD patients showed more prominent reductions in thalamic SC-FC coupling and glucose metabolism, which may represent a relatively specific neuroimaging alteration for BD. Notably, compared with glucose metabolic differences, SC-FC coupling abnormalities in the thalamus and ITG demonstrate greater intergroup differences, supporting the value of these features in the differential characterization of the two disorders. Furthermore, correlation analyses between SC-FC coupling/glucose metabolism and clinical parameters indicate significantly divergent patterns in associated brain regions between BD patients and MDD patients, which reflects distinct underlying neurobiological mechanisms. In the future, we will expand our sample cohort and build a long-term follow-up framework to further explore neuroimaging differences between adolescents with depressive episodes of BD and those with MDD, as well as their correlations with clinical symptoms. We aim to provide more reliable and visualizable neuroimaging evidence to support early accurate diagnosis and timely therapeutic intervention for these disorders.

Supplementary information

Below is the link to the electronic supplementary material.

Author contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Lu Liu, Fan Hu, Yanlei Huo and Xin Luo. The first draft of the manuscript was written by Simin Liu and Yingyan Zhong and all authors commented on previous versions of the manuscript. Chao Ma and Enzhao Cong read and approved the final manuscript.

Funding

This study was supported by the National Natural Science Foundation of China (Grant 82171974, 82372000), Tongji University 2024 Independent Fundamental Research Project: Multidimensional Cohort Study on “Anxiety Trait”-Endocrine-Amygdala-Hypothalamus Functional Imaging in Adolescents with Bipolar Disorder and Suicide Attempts (Project ID: 15012150111). Mechanism of Depression-Regulated Tumor Progression and Intervention Strategies Based on a Multidimensional Dynamic Cohort (Project ID: SYYYRH202507).

Declarations

Ethics approval

The authors indicate this study was in accordance with the ethical standards of the World Medical Association (Declaration of Helsinki), and the study received permission from the ethics committee of Tenth People’s Hospital of Tongji University (SHSY-IEC-5.0/24KY15/P01). This study was a cross-sectional study. The data are anonymous, and the requirement for informed consent was therefore waived.

Disclosure

The authors declare that there are no conflicts of interest.