Differences in biochemical metabolism and cognitive function between bipolar I and bipolar II disorder

Abstract

Background

This study aimed to characterise neurometabolic differences between bipolar disorder I (BD-I) and bipolar disorder II (BD-II), and to examine their associations with cognitive function.

Methods

A total of 50 patients diagnosed with BD-I, 80 patients with BD-II, and 50 healthy controls (HCs) were recruited for this study. Metabolite concentrations—specifically N-acetylaspartate (NAA) and choline-containing compounds (Cho)—were measured in the prefrontal white matter (PWM), anterior cingulate cortex (ACC), and thalamus through proton magnetic resonance spectroscopy (¹H-MRS). Cognitive performance was evaluated using the MATRICS Consensus Cognitive Battery (MCCB).

Results

When compared to HCs, patients with BD-II showed significantly higher Cho/Cr ratios in the right PWM and left ACC, along with lower NAA/Cr ratios in the right thalamus; compared to BD-II patients, those with BD-I exhibited higher Cho/Cr ratios in the right PWM and lower NAA/Cr ratios in the right thalamus; among BD-I patients, the Cho/Cr ratio in the left ACC was negatively correlated with measures of information processing speed and attentional vigilance.

Conclusions

This study demonstrates that BD-II patients exhibit greater cholinergic dysregulation in the left ACC and the right PWM, whereas BD-I patients show more pronounced neuronal dysfunction in the right thalamus. Furthermore, the left ACC Cho/Cr ratio was specifically associated with cognitive impairments in information processing speed and attentional vigilance, but only among patients with BD-I. This suggests that subtype-specific mechanisms underlie the relationship between neurometabolic abnormalities and cognitive deficits.

Clinical trial number

Not applicable.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12888-025-07697-0.

Introduction

Bipolar disorder (BD) is a serious and persistent mood disorder that significantly impairs patients’ social functioning and quality of life [29]. It ranks among the leading causes of disability worldwide [35]. According to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), Bipolar Disorder I (BD-I) is characterised by at least one manic episode, while Bipolar Disorder II (BD-II) requires at least one hypomanic episode and one major depressive episode [7]. Bipolar Disorder I (BD-I) is characterised by at least one manic episode, while Bipolar Disorder II (BD-II) requires at least one hypomanic episode and one major depressive episode [12]. During manic episodes, BD-I patients exhibit greater psychomotor agitation, impulsivity, irritability, and distractibility than BD-II patients [40]. In contrast, BD-II is associated with longer depressive episodes, more pronounced psychomotor retardation, and stronger suicidal ideation [12, 22]. However, early in the stages of illness, many patients display atypical or nonspecific symptoms, which are often intermittent and can be misdiagnosed as unipolar depression [37]. Consequently, diagnosing based solely on the recognition of manic or hypomanic symptoms carries a substantial risk of misdiagnosis.

Cognitive impairment can manifest early in bipolar disorder, primarily affecting attention, memory, and executive function [3]. Research on cognitive differences between subtypes remains limited; however, studies indicate specific impairments in motor speed, working memory, language acquisition, delayed memory, verbal fluency, and executive function across patients with BD I and BD II [13, 21], thereby supporting the notion of neurobiological distinctions between the two subtypes. Meta-analytical findings suggest that individuals with BD II experience cognitive deficits, albeit less severe than those observed in BD I [4]. Bipolar I disorder is characterised by reward hypersensitivity, which may lead to increased approach behaviours, tendencies toward hypomania or mania, emotional instability, and dysregulation [47]. Several studies have suggested that patients with BD-II demonstrate more significant impairments in executive function, verbal memory, and processing speed compared to those with BD I, likely attributable to more severe depressive symptoms in BD II patients [27]. Consequently, current research on neuropsychological abnormalities in BD I and BD II largely overlaps, although some studies report conflicting results.

Imaging techniques offer new evidence supporting the heterogeneity of bipolar disorder. Structural imaging reveals reduced volumes in regions such as the medial and ventral prefrontal cortex, anterior cingulate cortex (ACC), and insula in individuals with BD [19, 39, 48]. BD-I demonstrates notably significant reductions in grey matter volumes within the bilateral frontal, temporal, and occipital regions (Ha et al.). These structural alterations involve abnormal functional and anatomical interactions between the prefrontal cortex and subcortical structures responsible for self-related processing and the regulation of internal and external demands [14]. Similar deficits were not observed in BD-II [9]. Conversely, patients with BD-II displayed significant atrophy of the ACC and increased surface area of the right insula [9, 49], in addition to an increased volume of the left caudate nucleus [8]. Notably, the caudate nucleus exhibits significantly stronger positive functional connectivity with the left ventral striatum during reward anticipation [8]. Moreover, BD-II patients demonstrate more pronounced ventral striatal activity and larger left caudate volume during reward anticipation [8]. Diffusion tensor imaging identified fiber damage within the thalamus, ACC, and subfrontal regions across both subtypes, with additional fiber alterations in the temporal and subfrontal regions observed in BD-II. Fractional anisotropy (FA) values in the right inferior frontal gyrus for BD-I and the left middle temporal gyrus for BD-II correlated with measures of executive function(J.-X. Liu et al., [25]. Additionally, FA values in the left medial temporal lobe and inferior frontal lobe were associated with Young Mania Rating Scale (YMRS) and hypomanic episode scores, respectively. Consequently, differences in structural and functional connectivity between BD I and BD II may reflect distinct patterns of cognitive function across various brain regions.

Researchers have continually examined stable and effective biomarkers to improve the diagnosis and treatment of BD I versus BD II. Proton magnetic resonance spectroscopy (¹H-MRS) provides a non-invasive technique for measuring N-acetylaspartate (NAA) and choline-containing compounds (Cho). Unlike less detailed volumetric estimates,¹H-MRS provides deeper insights into neural abnormalities at the cellular and metabolic levels [11]. N-acetylaspartate (NAA), a highly concentrated brain metabolite, is synthesized in neuronal mitochondria from L-aspartate and acetyl-CoA, and its production depends on energy metabolism. Mitochondria influence neuroplasticity, development, and axonal repair, with NAA levels serving as indicators of mitochondrial energy health [30]. Choline, derived from glycerophosphocholine and phosphatidylcholine, is related to membrane integrity and myelin formation [30]. The breakdown of cellular membranes, reflected by higher choline levels, results in the release of choline-containing compounds [43]. A recent meta-analysis found region-specific changes in NAA levels in bipolar depression, showing decreases in the left prefrontal white matter (PWM) and increases in the left dorsolateral prefrontal cortex (DLPFC), in contrast to stable levels in the ACC and hippocampus [10]. A separate meta-analysis revealed that patients with BD have a significantly higher choline-to-creatine (Cho/Cr) ratio in the frontotemporal cortex [38]. In our prior research, patients with bipolar disorder exhibited reduced N-acetylaspartate to creatine(NAA/Cr) ratios in the PWM and ACC [10, 26, 38], as well as increased NAA/Cr ratios in the left thalamus [52]. Nonetheless, earlier investigations did not identify significant differences between BD-I and BD-II.

Therefore, this study aims to employ ¹H-MRS technology to systematically compare neurometabolite levels in various brain regions, such as the ACC, PWM and thalamus, between patients with BD-I and BD-II. By integrating these findings with assessments of cognitive function and clinical symptoms, we seek to elucidate the neurochemical distinctions between the two subtypes. This research aims to provide novel imaging evidence to enhance classification accuracy and deepen understanding of bipolar disorder.

Methods

Participants

Patients with BD

This study took place in the Department of Psychiatry at the First Affiliated Hospital of Central South University in Guangzhou, China. The criteria for including bipolar disorder patients were: (1) diagnosis of BD based on the DSM-5; (2) first-time presentation without previous psychiatric treatment; (3) aged 18 to 60; (4) a 24-item Hamilton Depression Rating Scale(HDRS-24) score of 20 or more, and a YMRS score below 6 [50]; (5) Han Chinese ethnicity and right-handedness; and (6) voluntary participation with written informed consent. Exclusion criteria were: (1) other psychiatric disorders; (2) several primary neurodevelopmental disorders, such as attention-deficit/hyperactivity disorder (ADHD) and autism spectrum disorder, were considered comorbid conditions; (3) current or past neurological disorders (e.g., epilepsy, sequelae of traumatic brain injury) or severe systemic diseases; (4) substance abuse history; (5) prior use of psychotropic drugs, psychotherapy, or electroconvulsive therapy; (6) pregnancy or breastfeeding; and (7) MRI contraindications. A total of 130 eligible BD patients were enrolled, with blood samples, cognitive tests, and neuroimaging data collected.

Healthy controls

Healthy controls (HCs) were recruited via community and hospital advertisements and matched to the patient group by age. All potential controls underwent clinical interviews to ensure they had no personal or first-degree family history of psychiatric disorders. To be eligible, HCs had to score less than 8 on the HDRS-24 and less than 6 on the YMRS. The exclusion criteria were the same as those for the MDD group. Following screening, 50 qualified HCs were enrolled in the study.

The Ethics Committee of the First Affiliated Hospital of Jinan University authorised the study protocol. All participants gave voluntary consent after being thoroughly informed about the study procedures through both verbal and written explanations. (Fig. 1).

Fig. 1.

Study enrollment flowchart

Clinical characteristics and subgroup

Baseline assessments included collecting demographic data and clinical histories through structured interviews. Additionally, a retrospective review of medical records confirmed past episodes of mania or hypomania, supporting the diagnosis of BD-I or BD-II (Fig. 1).

Cognitive function assessment

A cognitive assessment was conducted using the MCCB. For both HC and patients with BD, the established clinical validity and test-retest reliability have been maintained. The administration of the full MCCB, which lasts approximately 70 min, was employed to evaluate seven domains: processing speed, attention/vigilance, working memory, verbal learning, visual learning, reasoning/problem-solving, and social cognition, in addition to a comprehensive global composite score [33]. The assessment comprised the Trail Making Test Part A (TMT-A); Brief Assessment of Cognition in Schizophrenia: Symbol Coding; Hopkins Verbal Learning Test (HVLT); Wechsler Memory Scale: Spatial Span; Neuropsychological Assessment Battery (NAB): Mazes; Brief Visuospatial Memory Test; Category Fluency; Mayer-Salovey-Caruso Emotional Intelligence Test (MSCEIT): Managing Emotions; and the Continuous Performance Test: Identical Pairs [34].

MRI data acquisition and preprocessing

A total of 130 medication-naïve patients with BD and 50 HCs underwent magnetic resonance imaging (MRI) and ¹H-MRS scans. A GE Discovery MR 750 system operating at 3.0 T (GE Healthcare, Milwaukee, WI), equipped with an 8-channel head coil, was used for data collection. Participants were positioned supine with the nasion as the anatomical landmark. Earplugs and foam padding were used to reduce acoustic noise and limit head movement.

To screen for structural brain abnormalities, the following sequences were acquired: axial T1-weighted FLAIR (TR/TE = 1750/24 ms) and fast spin-echo T2-weighted imaging (TR/TE = 8400/145) ms). A Point-Resolved Spectroscopy (PRESS) sequence combined with CHESS-based water suppression was used to obtain 2D multi-voxel ¹H-MRS data for spectroscopic analysis. Acquisition parameters included: TR = 1000 ms, TE = 144 ms, spatial matrix = 18 × 18 cm, field of view = 240 × 240 mm, slice thickness = 10 mm, and nominal voxel size = 7.5 × 7.5 × 10 mm³. Saturation bands were positioned around the volume of interest (VOI) to minimise lipid contamination from subcutaneous tissues.

Automated pre-scanning ensured that the full width at half maximum (FWHM) stayed under 10 Hz and water suppression efficiency exceeded 98%. Spectra that did not meet these quality standards were excluded. VOIs were sequentially positioned in the PFC, ACC, and thalamus (Figs. 1 and 2). Each VOI acquisition lasted 5 min and 28 s, leading to a total scan time of approximately 10 min and 56 s per participant for the ¹H-MRS protocol.

Fig. 2.

Magnetic resonance image (MRI) scan of a healthy control subject showing the location of magnetic resonance spectroscopy (MRS) of the volume of interest (VOI) placed in the left and right PFC, ACC, and thalamus. The large white boxes represent the VOIx for MRS acquisition, and two small boxes depict the individual VOIs of bilateral PFC, ACC, and thalamus for spectral analysis. For each subject, the ROI size were found to be identical in the left and right PFC, ACC, and thalamus using the mirror symmetry tools from the Functool software. Note: NAA, N-acetyl aspartate; Cho, choline containg compounds; Cr, creatine

All spectral data were analysed using the GE Advantage Workstation (AW 4.2_07 FuncTool). An experienced radiologist, blinded to the participants’ diagnoses, designated voxels for spectroscopic analysis. Spectra were acquired from voxels repositioned within the bilateral PFC, ACC, and thalamus. Each volume of interest (VOI) comprised four individual voxels (1 × 1 × 1 mm³), and the final metabolite value for each VOI was determined as the mean of these four voxels. Spectral analysis encompassed the quantification of peak areas corresponding to choline (Cho, 3.22 ppm), creatine (Cr, 3.03 ppm), and N-acetylaspartate (NAA, 2.02 ppm). Biochemical neurometabolic alterations were evaluated through the ratios NAA/Cr and Cho/Cr obtained from the bilateral PFC, ACC, and thalamic regions(Fig. 2).

Statistical analysis

Statistical analyses were conducted utilising SPSS Statistics (version 20.0; SPSS Inc., Chicago, IL, USA) and R (version 3.3.1). The normality of all variables (e.g., demographics, cognitive scores, and neurometabolic ratios) was assessed using the Kolmogorov-Smirnov test, while Levene’s test evaluated the homogeneity of variances. A significance level of p < 0.05 was used for both tails throughout the analysis.

Differences in demographics, clinical variables, biochemical metabolite ratios, and cognitive indices across groups were examined using one-way ANOVA, accompanied by Bonferroni post hoc analyses for variables that followed a normal distribution. For variables that did not adhere to normality, the Kruskal–Wallis test with Bonferroni adjustments was employed. Categorical variables, such as gender, were analysed using chi-square tests. Furthermore, group comparisons for clinical variables were conducted according to data distribution: the two-sample t-test was used for normally distributed data, while the Mann-Whitney U test was applied to non-normal data. All p-values were then adjusted for multiple comparisons using the Bonferroni correction.

The Spearman/Pearson correlation test examined relationships among biochemical metabolite ratios, cognitive function, and clinical features within the BD I and BD II groups, with Bonferroni correction for multiple comparisons.

Results

Demographic and clinical information

The study encompassed a sample of 130 medication-naïve individuals diagnosed with bipolar disorder, consisting of 50 patients with BD-I (16 males, representing 32.0%, and 34 females, representing 68.0%) and 80 patients with BD-II (19 males, representing 23.8%, and 61 females, representing 76.2%). The baseline demographic and clinical profiles of the study cohort are presented in Table 1. The mean age was 24.18 ± 6.74 years for BD-I patients, 23.13 ± 4.61 years for BD-II patients, and 23.52 ± 5.48 years for HC. There were no statistically significant differences in age, gender distribution, or years of education among the three groups (all p > 0.05). Likewise, there were no notable differences between BD subtypes in key clinical measures such as age at onset, illness duration, number of episodes, and severity scores (HDRS-24/YMRS). As anticipated, both patient groups exhibited significantly higher HDRS-24 and YMRS scores compared to healthy controls (p < 0.001).

Table 1.

Demographics and clinical data of all participants (mean ± SD)

	BD I group	BD II group	HCs	F, X²,Z	P, P	p value e
Number of subjects	50	80	50	-	-
Age (year)	24.18 ± 6.736	23.13 ± 4.610	23.52 ± 5.481	1.195	0.054 a
Sex (male/female)	16/34	19/61	22/28	-	-
yearseducation	14.32 ± 2.369	14.506 ± 2.438	15.12 ± 3.068	1.323	0.269 a
Age of onset (year)	19.65 ± 7.429	19.24 ± 6.053	-	0.137	0.712 b
Duration of illness (month)	54.33 ± 48.601	48.601 ± 50.097	-	0.239	0.626 b
Number of fredepression Number of fremania/ hypomanic	2.00 [1.25,2.00] 1.00 [1.00,2.00]	1.00 [2.00,2.00] 1.00 [1.00,1.00]	-	-0.920 -0.714	0.357 c 0.475 c
24-item HDRS score	25.81 ± 5.270	25.19 ± 4.368	1.88 ± 3.353	563.22	< 0.001 a***	BD I, BD II > HCs d***
YMRS score	2.00 ± 2.053	2.39 ± 2.027	0.42 ± 1.993	18.575	< 0.001 a***	BD I, BD II > HCs d***

BD I, bipolar I disorder; BD II, bipolar II disorder; HCs, healthy controls;24-item HDRS, Hamilton Depression Rating Scale; YMRS, Young Mania Rating Scale; ^a One-Way ANOVA analysis; ^b X² test;^c Bonferroni post hoc test;^c Mann-Whitney U test; p<0.05*; p<0.005**; p<0.001***

Biochemical metabolite ratios

Table 2 shows the regional NAA/Cr and Cho/Cr ratios for the BD-I, BD-II, and HC groups. Several significant differences were found: BD-II patients had higher Cho/Cr ratios in the right PWM and left ACC compared to healthy controls, while BD-I patients had lower NAA/Cr ratios in the right thalamus. In direct comparisons, BD-II patients showed a significantly higher Cho/Cr ratio in the right PWM (p = 0.007) and a lower NAA/Cr ratio in the right thalamus (p = 0.038) than BD-I patients (see Fig. 3).

Table 2.

Comparisons of biochemical metabolite ratios among BD I, BD II and HCs group (mean ± SD)

	BD I group	BD II group	HCs	F, X²,Z	p	p value b
Left PWM
NAA/Cr	2.13 ± 0.43	2.02 ± 0.43	2.11 ± 0.39	1.66	0.194 a
Cho/Cr	1.16 ± 0.20	1.20 ± 0.21	1.12 ± 0.20	1.78	0.172 a
Right PWM
NAA/Cr	2.11 ± 0.39	2.22 ± 0.37	2.21 ± 0.27	1.6	0.205 a
Cho/Cr	1.07 ± 0.21	1.18 ± 0.23	1.08 ± 1.77	5.18	0.007 a**	BD II > BD I, HC b*
Left ACC
NAA/Cr	1.84 ± 0.3	1.83 ± 0.27	1.80 ± 0.22	0.19	0.831 a
Cho/Cr	1.10 ± 0.17	1.14 ± 0.19	1.04 ± 0.17	4.79	0.009 a**	BD II > BD I, HC b*
Right ACC
NAA/Cr	1.75 ± 0.26	1.80 ± 0.22	1.84 ± 0.22	2.1	0.126 a
Cho/Cr	1.08 ± 0.16	1.07 ± 0.18	1.10 ± 0.16	0.47	0.626 a
Left thalamus
NAA/Cr	2.07 ± 0.34	2.13 ± 0.31	2. 11 ± 0.31	0.58	0.563 a
Cho/Cr	0.90 ± 0.18	0.91.±0.0.19	0.94 ± 0.21	0.288	0.750 a
Right thalamus
NAA/Cr	1.92 ± 0.35	2. 07 ± 0.32	2. 09 ± 0.26	4.21	0.013 a*	BD II, HC > BD I b*
Cho/Cr	0.85 ± 1.56	0.87 ± 0.16	0.90 ± 0.20	0.99	0.373 a

BD I, bipolar I disorder; BD II, bipolar II disorder; HCs, healthy controls;24-item HDRS, Hamilton Depression Rating Scale; YMRS, Young Mania Rating Scale; NAA, N-acetylaspartate; Cho, choline-containing compounds; Cr, Crcreatine; ^a One-Way ANOVA analysis; ^b Bonferroni post hoc test; p<0.05*; p<0.005**; p<0.001***

Fig. 3.

Comparisons of biochemical metabolite ratios among BD I, BD II, and HCs group (mean ± SD). Abbreviations: BD I, bipolar I disorder; BD II, bipolar II disorder; HCs, healthy controls;24-item HDRS, Hamilton Depression Rating Scale; YMRS, Young Mania Rating Scale; a One-Way ANOVA analysis; b Bonferroni post hoc test; p<0.05*; p<0.005**; p<0.001***

Cognitive function indices

Table 3 summarizes cognitive performance across BD-I, BD-II, and HCs groups as assessed by the MCCB. Compared to HCs, both BD-I and BD-II patients showed significant impairments in processing speed, attention-vigilance, verbal learning, visual learning, reasoning and problem-solving, social cognition, and the overall composite score (all p < 0.05). However, there were no significant differences in MCCB domain scores between BD-I and BD-II patients (all p > 0.05) (Table 3).

Table 3.

Comparisons of cognitive function indices among BD I, BD II and HCs group (mean ± SD)

	BD I group	BD II group	HCs	F, X², Z	p value	p value e
Speed of processing	45.84 ± 10.68	45.24 ± 8.25	51.73 ± 6.79	9.344	< 0.001 a***	HCs > BD I, BD II b*
Attention vigilance	44.82 ± 8.41	45.87 ± 9.16	49.45 ± 7.51	3.984	0.020 a**	HCs > BD I, BD II b*
Working memory	47.52 ± 10.97	46.37 ± 11.19	50.02 ± 9.49	1.746	0.178 a
Verbal learning	43.66 ± 7.92	45.83 ± 9.65	53.16 ± 7.34	16.461	< 0.001 a***	HCs > BD I, BD II b***
Visual learning	45.89 ± 7.74	43.96 ± 9.64	51.82 ± 7.22	12.944	< 0.001 a***	HCs > BD I, BD II b***
Reasoning problem solving	43.57 ± 11.68	45.16 ± 11.79	52.02 ± 8.91	8.252	< 0.001 a***	HCs > BD I, BD II b***
Social cognition	47.57 ± 10.97	46.53 ± 10.56	51.45 ± 9.95	3.369	0.037 a***	HCs > BD I, BD II b*
Composite	43.09 ± 8.38	43.07 ± 8.50	51.86 ± 6.33	21.484	< 0.001 a***	HCs > BD I, BD II b***

BD I, bipolar I disorder; BD II, bipolar II disorder; HCs, healthy controls; ^a One-Way ANOVA analysis; ^b Bonferroni post hoc test; p<0.05*; p<0.005**; p<0.001***

Correlation results

Figure 2 demonstrates the associations between neurometabolic ratios and cognitive scores within the BD-I cohort. In individuals diagnosed with BD-I, Cho/Cr ratio in the left ACC exhibited significant negative correlations with processing speed (r = − 0.321, p = 0.034; Fig. 2A) and attention vigilance (r = − 0.302, p = 0.046; Fig. 2B). Conversely, in the BD-II group (Fig. 3), no significant relationships were observed between neurometabolic ratios and cognitive performance metrics. Nevertheless, among healthy controls, a positive correlation was identified between the left ACC Cho/Cr ratio and visual learning (r = 0.308, p = 0.031; Fig. 3A), as well as between the right PFC Cho/Cr ratio and verbal learning (r = 0.293, p = 0.041; Fig. 3C).

Moreover, both patient groups showed no significant correlation between the assessed clinical features (including illness chronology, episode frequency, and depression severity) and their neurometabolic or cognitive profiles (Figs. 4 and 5).

Fig. 4.

The correlations among abnormal biochemical metabolite ratios and cognitive indices in HCs. Note: Cho, choline containg compounds; Cr, creatine; ACC, anterior cingulate cortex;

Discussion

This study aimed to compare neurometabolic profiles between patients with BD-I and BD-II and to examine the relationship between key cerebral metabolites and cognitive performance across bipolar disorder subtypes. The principal findings are as follows: (1) When compared to HC, patients with BD-II demonstrated significantly higher ratios of Cho/Cr in the right PWM and left ACC, along with lower NAA/Cr ratios in the right thalamus; (2) When compared to BD-II patients, those with BD-I exhibited higher Cho/Cr ratios in the right PWM and lower NAA/Cr ratios in the right thalamus; (3) Among BD-I) patients, the Cho/Cr ratio in the left anterior cingulate cortex (ACC) correlates with reduced performance in processing speed and attentional vigilance.

This study identified a significant increase in the Cho/Cr ratio in the right PWM among patients with BD-II in comparison to those with BD-I. Choline, a fundamental component of membrane phospholipids, is particularly abundant in oligodendrocytes [46]. Elevated Cho levels primarily indicate accelerated phospholipid turnover or membrane degradation, releasing choline-containing compounds and implying impaired neuronal phospholipid metabolism [15]—a mechanism associated with bipolar disorder. A systematic review also affirmed that Cho levels tend to be elevated in patients with bipolar disorder [43], especially within the prefrontal cortex and bilateral ventrolateral prefrontal regions [36, 42]. Furthermore, a comparative study observed increased Cho levels specifically in the left ACC and left PWM of bipolar depression patients relative to unipolar depression patients [23], suggesting that PWM Cho alterations may serve as a neurochemical marker specific to a bipolar disorder subtype. Our study confirms previous findings of increased Cho/Cr ratios in the right PWM, a characteristic observed in both BD subtypes compared with healthy controls. Importantly, we observed that BD-II patients exhibit even higher ratios than those of BD-I patients in this region. Given that earlier research associates greater improvement in depressive symptoms with lower Cho/Cr ratios [31], our results imply that BD-II involves more profound membrane metabolic dysregulation in the right PWM. This provides a plausible neurobiological foundation for the distinct clinical features and disease progression patterns observed between the two BD subtypes.

Compared to HCs, patients diagnosed with BD-II exhibited a significantly elevated Cho/Cr ratio within the left ACC. This observation is corroborated by a meta-analysis examining neurochemical metabolism in the ACC of individuals with bipolar depression, which similarly identified increased ACC Cho/Cr levels [38]. An additional meta-analysis, characterized by larger effect sizes and reduced heterogeneity, further implies that heightened ACC Cho levels could serve as a potential trait marker for bipolar disorder, as this elevation has been documented in euthymic, depressive, and medication-free individuals, but not during manic episodes [38, 43]. Accordingly, the increased Cho/Cr ratio in the left ACC of BD-II patients may function as a stable and prospective trait biomarker for this condition. Place this neurochemical finding within a broader neurobiological framework; numerous neuroimaging investigations, including extensive analyses conducted by the ENIGMA consortium, have consistently identified cortical thinning in frontal regions—particularly the ACC—as one of the most prominent structural abnormalities associated with bipolar disorder [18, 19, 44]. From a molecular perspective, Cho signal detected via ¹H-MRS primarily originates from phosphocholine (PChol) and glycerophosphocholine (GPC). Phosphocholine serves as a crucial precursor in the biosynthesis of membrane phospholipids, while both phosphocholine and glycerophosphocholine are released during their breakdown [5, 16]. Consequently, the intensity of the Cho signal depends on cell membrane density and the rate of membrane and myelin turnover. Enhanced breakdown of membrane phospholipids, particularly in myelin, leads to an increased ¹H-MRS choline signal due to the release of choline-containing compounds [5, 16]. Our findings provide a potential molecular explanation for these structural alterations; elevated Cho levels reflect ongoing disturbances in neuronal membrane phospholipid metabolism within the ACC, such as increased membrane turnover or breakdown. This persistent metabolic imbalance could compromise the integrity of neuronal and glial cell membranes, disrupting synaptic remodeling processes and ultimately contributing to cortical thinning, which is observable at a macroscopic level.

Our objective was to establish a connection between cerebral metabolic alterations and clinical symptoms. In patients diagnosed with BD-I, the Cho/Cr ratio in the left ACC demonstrated significant negative correlations with performance in specific cognitive domains, including processing speed and attentional vigilance. This indicates that a higher Cho/Cr ratio in the left ACC is associated with diminished performance on these cognitive assessments. Impaired sustained attention constitutes one of the most prevalent cognitive deficits observed in bipolar disorder [24]. The ACC functions as a crucial connector between the prefrontal cortex and the limbic system, facilitating the integration of cognitive and emotional processes [38]. Functional abnormalities within the ACC of individuals with BD may contribute to emotional dysregulation, thereby affecting cognitive flexibility [1]. Prior studies have also reported abnormal interactions between the ACC and the default mode network (DMN) in BD, which may underlie executive dysfunction. Research involving adolescent BD populations has similarly demonstrated a reduction in ACC thickness, implying that such early developmental modifications may be associated with cognitive deficits [45]. Our investigation offers convergent evidence at the metabolic level, thereby reinforcing the hypothesis that dysfunction within the ACC constitutes a neurobiological foundation for the fundamental cognitive impairments observed in BD-I patients. These findings suggest cholinergic metabolic irregularities in the left ACC as a potential source of disrupting the cognitive-emotional integration circuitry, ultimately compromising information processing efficiency.

However, no significant correlation was identified between the left ACC Cho/Cr ratio and the specified cognitive domains in patients with BD-II. This discrepancy suggests the potential existence of distinct neurobiological mechanisms among different bipolar disorder subtypes. Our findings are consistent with prior research focused on BD-I: one study demonstrated persistently higher ACC Cho/Cr ratios in BD-I patients during depressive or euthymic states [41], whereas another reported significantly elevated levels of choline-containing compounds (GPC + PC)—which are associated with membrane metabolism—in the ACC of BD-I patients exhibiting rapid cycling features [6]. Glycerophosphocholine (GPC) and phosphocholine (PC) are crucial metabolites involved in the breakdown and synthesis of membrane phospholipids, respectively [28]. Consequently, increased Cho levels may indicate heightened disease activity. In BD-I, ongoing disturbances in membrane metabolism or inflammatory processes are likely to induce local neuronal dysfunction or early tissue damage, thereby directly contributing to cognitive deficits, particularly in information processing speed and attentional vigilance.

The Thalamic NAA/Cr ratios showed a gradual decrease across the groups. This ratio was significantly lower in both BD-I and BD-II patients compared to HCs, with BD-I patients exhibiting a more considerable reduction than their BD-II counterparts.NAA is primarily localized within neurons, and diminished NAA levels are generally regarded as an indicator of neuronal dysfunction or loss [30]. These findings correspond with earlier structural neuroimaging research, which has documented thalamic volume reduction in patients with BD-I [32]. The thalamus, a critical subcortical structure, maintains bidirectional connections with the prefrontal cortex [2] and has been implicated in the pathophysiology of bipolar disorder. Notably, such thalamic volumetric impairments have been identified even during euthymic phases [17], implying that thalamic abnormality may constitute a persistent trait-like characteristic in BD-I. The literature reports mixed results, with some studies showing increased or unchanged thalamic NAA levels in BD-I patients [17, 20], with other studies documenting significantly higher NAA/Cr ratios specifically in the left thalamus of those diagnosed with BD-II. These differences may originate from variations in clinical characteristics, such as the stage of illness or medication history, or from methodological discrepancies across studies. The thalamus, functioning as a central hub for sensory integration and transmission, plays a crucial role; dysfunction within this structure can directly interfere with prefrontal-limbic circuits responsible for emotional regulation. Alterations in functional connectivity within the thalamocortical pathways, particularly between the thalamus and the ACC, have been consistently documented in bipolar disorder [51]. Therefore, we hypothesize that the more pronounced reduction in thalamic NAA/Cr observed in BD-I patients may contribute to their more severe manic episodes and emotional instability by disrupting the integrity of these neural pathways.

A major strength of this study is its groundbreaking examination of neurobiological differences between BD-I and BD-II, with an emphasis on neurometabolic profiles and cognitive performance. These results could aid in distinguishing the subtypes clinically and serve as a basis for exploring their unique neurophysiological mechanisms. Additionally, we investigated potential correlations between plasma metabolite levels and cognitive abilities. Another key advantage is that all bipolar disorder patients involved were medication-naïve, which reduces the confounding impact of psychotropic drugs on the findings.

Several limitations of this study should be acknowledged. First, its cross-sectional design, with all participants assessed during depressive episodes, means that mood state may have influenced both metabolite levels and cognitive performance. Future longitudinal studies tracking the same patients across different mood states are needed to determine whether the observed metabolic alterations serve as state markers or stable endophenotypes. Additionally, a conceptual limitation is the insufficient integration of comorbid conditions and transdiagnostic approaches. BD shows substantial symptomatic overlap and high comorbidity with conditions such as ADHD and anxiety disorders. The observed cognitive deficits—including impaired attention and reduced information processing speed—and prefrontal-limbic metabolic abnormalities may partly reflect general psychopathological features that transcend diagnostic boundaries. Future research should systematically assess and account for comorbid conditions or directly compare findings with those from other psychiatric patient groups.

Conclusions

This study provides novel evidence of neurobiological distinctions between BD-I and BD-II by integrating neurometabolic, cognitive, and clinical profiles. It demonstrates that BD-II patients exhibit greater cholinergic dysregulation in the right prefrontal white matter and left ACC, whereas BD-I patients show more pronounced neuronal dysfunction in the right thalamus. Furthermore, the left ACC Cho/Cr ratio was specifically associated with cognitive impairments in information processing speed and attentional vigilance, but only among patients with BD-I. This suggests that subtype-specific mechanisms underlie the relationship between neurometabolic abnormalities and cognitive deficits. Although exploratory, this research underscores the value of ¹H-MRS in elucidating subtype-specific pathophysiological processes. Subsequent studies involving larger, longitudinal cohorts and comprehensive neuropsychological assessments are essential to validate these findings and to enhance understanding of how metabolic dysregulation, structural variations, and clinical progression interact in bipolar disorder.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Concept and design: Shuming Zhong and Yanbin Jia. Data analysis: Rongxu Zhang and Dong Huang. Drafting the manuscript: Rongxu Zhang and Dong Huang. Revising the manuscript: Shuming Zhong and Yanbin Jia. Providing data: Shunkai Lai, Ying Wang, Yiliang Zhang, Jiali He, Guanmao Chen, Shuya Yan, Pan Chen, Xiaodan Lu, Xiaosi Huang, and Yanbin Jia.

Funding

This work was supported by the National Natural Science Foundation of China (no. 82271564), the National Key R&D Program of China (2022YFB4500600), the National Natural Science Foundation of China, China (no. 82271564), and the Science and Technology Projects of Guangzhou(202206060001). The founders have not played any roles in study design, data collection, analysis, manuscript writing, and the decision to publish.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethical approval

This study was conducted in accordance with the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines. The research protocol was approved by the Ethics Committee of the First Affiliated Hospital of Jinan University, Guangzhou, China. All subjects signed a written informed consent form. Not all or part of this information is published elsewhere; the manuscript has not yet been considered for publication in any other journal; all the authors have personally and actively participated in the actual work of the manuscript and will bear the common and individual responsibility for their content.

Competing interests

The authors declare no competing interests.