Bipolar disorder at mixed states and major depressive disorder with mixed features differ in peripheral biochemical parameters

Abstract

Background

Little is known about the peripheral biochemicals between bipolar disorder at mixed episodes (BDM) and major depressive disorder with mixed features (MDM). This retrospective study was aimed to compare the peripheral biochemical parameters between patients with BDM and MDM.

Methods

This study included data from 269 BDM patients and 86 MDM patients. Biochemical markers covering immune-inflammatory, liver function, metabolic, and thyroid hormone indices were analyzed. Logistic regression models were employed to evaluate associations between biochemical markers and diagnosis. Network analysis and Principal Component Analysis (PCA) was also performed to investigate the relationships among these parameters.

Results

BDM patients had higher neutrophil percentage (NEUT%), white blood cell count (WBC), free triiodothyronine (FT3) and free thyroxine (FT4), while MDM patients exhibited higher levels of C-reactive protein (CRP), direct bilirubin (DBIL) and prealbumin (PA). NEUT%, WBC, FT3 and FT4 showed positive association with the diagnosis of BDM, while PA and DBIL displayed negative correlation with BDM. No significant differences in either network structure or global strength were found between BDM and MDM groups.

Conclusion

Peripheral biochemical markers, particularly those related to the immune-inflammatory factors and thyroid hormones, differ between BDM and MDM, which could contribute to better understanding of potential status mechanism under the disorders.

Trial registration

International Clinical Trials Registry Platform: NCT03949218. Registered on 13/05/2019.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12888-025-06800-9.

Introduction

Bipolar disorder (BD), characterized by recurrent episodes of elevated mood and depression, has been reported to have an overall lifetime prevalence of 2.4% [1]. In the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), the mixed states of BD were defined as the presence of opposite polarity symptoms during a manic or depressive episode [2]. In the STEP-BD study, Goldberg et al. reported that during a major depressive episode, only 31.2% of BD patients had no manic symptoms [3]. A Previous study concluded that the average duration of mixed episode of BD was about 15.5 weeks [4]. In DSM-5, the manic symptoms below the threshold for hypomania (mixed features) in major depressive disorder (MDD) was recognized as a structural bridge between BD and MDD for the first time [5]. A national survey in the United States reported that the prevalence of MDD patients with mixed features (MDM) was 15.5% [6].

The current criteria for MDM are defined as meeting depressive episode criteria, and three of the following seven symptoms in the last 2 weeks at least every day for a noticeable amount of time to some degree: (1) Elevated or expansive mood; (2) Inflated self-esteem or grandiosity; (3) More talkative than usual or pressure to keep talking; (4) Flight of ideas or subjective sense that thoughts are racing; (5) Increase in energy or goal-directed activity; (6) Increase or excessive involvement in activities that have a high potential for painful consequences; (7) Decreased need for sleep [7]. Clinical characteristics of BD patients at mixed episodes (BDM) mainly consist of three parts: (1) Manic symptoms: greater mood lability, irritability, grandiosity, euphoria, pressured speech and decreased need for sleep [3, 8]; (2) Depressive symptoms: dysphoric mood, anxiety, excessive guilt and suicidality [9]; (3) Non-mood symptoms: anxiety, agitation and psychosis [10, 11]. Some symptoms of MDD patients, such as agitation and diminished concentration, could be similar to those of mania [12]. Overlapping symptoms between MDM and BDM might hinder the accurate diagnosis and treatment of disease, especially for patients at depressive episode with mixed features [13].

The World Federation of Societies of Biological Psychiatry (WFSBP), Canadian Network for Mood and Anxiety Treatments (CANMAT) and International Society for Bipolar Disorders (ISBD) recently published guidelines for management of BD patients at mixed states. The researchers found manic symptoms in bipolar mixed states appeared responsive to atypical antipsychotics, such as olanzapine; for depressive symptoms, addition of ziprasidone may be beneficial; and for recurrence prevention, olanzapine, quetiapine, valproate and lithium should be considered [14]. Existing evidence suggests that atypical antipsychotics and divalproex are beneficial for treating acute mixed presentations of BD [15, 16]. Although evidence-based data supports are limited, antidepressants should be avoided in bipolar mixed states in general [17]. Unlike BD at mixed states, there are currently no treatment algorithm for MDD with mixed features in comparison with those without mixed features [18]. MDM may be more treatment resistant, which requires augmentation with the second generational antipsychotics or mood stabilizing agent. Atypical antipsychotics, such as asenapine, lurasidone, olanzapine, quetiapine and ziprasidone, have shown some efficacy for the treatment of depression with mixed features [19–21].

A growing body of research implicates immune-inflammatory processes in the pathophysiology of mood disorders, with peripheral biochemical markers offering potential insights into their underlying mechanisms. C-reactive protein (CRP), an acute-phase protein produced by the liver in response to inflammation, has emerged as a key indicator across psychiatric and somatic conditions. In MDD, elevated CRP levels are well-documented and associated with systemic inflammation, contributing to depressive symptom severity [22]. Similarly, in BD, CRP concentrations are increased, with meta-analyses showing higher levels during manic episodes compared to euthymic or depressive states [23]. Comparative studies between BD and MDD further suggest that CRP may be more elevated in BD, particularly during mania, than in unipolar depression [24]. However, the role of CRP in mixed states, where manic and depressive symptoms coexist, remains underexplored, and its differential expression in BDM versus MDM could illuminate distinct inflammatory profiles. CRP’s association with inflammation extends beyond mood disorders to a wide array of systemic conditions, highlighting its broad clinical relevance and complexity. Elevated CRP levels have been reported in diabetic nephropathy [25], where it correlates with disease progression, and in autoimmune thyroiditis [26], reflecting inflammatory activity in endocrine dysfunction [27]. Similarly, CRP is implicated in hepatitis [28], underscoring its sensitivity to metabolic and hepatic inflammation. Moreover, recent evidence links CRP-based inflammatory markers to COVID-19 mortality, with higher levels predicting poorer outcomes [29]. These findings across diverse conditions suggest that CRP is a non-specific marker of systemic inflammation, modulated by disease-specific contexts, which may parallel its variable expression in mood disorders.

Abundant studies have been trying to find out objective differences between BD and MDD, especially for BD patient with onset as depressive episode [30, 31]. However, studies on differentiating MDD patients with mixed features from those at mixed states in BD are rare. Developing biological data for spectrum of mood disorders are needed, for the reason that categories in terms of symptoms by DSM-5 and other similar systems of diagnosis may not reflect what the brain and the body is experiencing [32]. Since it’s still under debate that should MDD with mixed features be considered a unique episode or as the path to developing bipolar disorder [7], it’s important to find biological, neuroimaging or genetic basis of MDD with mixed features for targeted treatment. Previous studies have found out significant differences of biological biomarkers, such as inflammatory factors and metabolic indexes, between BD and MDD [33]. Whether there are differences of peripheral biomarkers between BDM and MDM, as well as the biological mechanisms of the diseases are unknown. In the present study, biochemical from routine hematological examination of BDM and MDM patients were utilized to explore the biological mechanism under the development of the two subtypes of mood disorders. These biochemical parameters consisted of five parts: (i) immune-inflammatory (8 indexes); (ii) liver function (7 indexes); (iii) metabolism (6 indexes); (iv) thyroid hormones (5 indexes).

Methods

Source of data

This is a retrospective, cross sectional study. The data gathered for this study was from the Information and Statistics Department of Shanghai Mental Health Center (SMHC) from 1st January 2008 to 31st December 2018, as previously reported [34, 35]. Information of 4473 records of inpatients with mood disorders were obtained. All the patients meet diagnosis of ICD-10 F31 (bipolar disorder) or F32 (depressive disorder and its subtypes). Diagnosis was conducted and confirmed by three levels of doctors including at least one attending physician and one chief physician in psychiatry. Information of these patients included medical history and clinical biochemical data for the first measurement at admission. The fasting venous blood was collected between 7:00 a.m. and 8:00 a.m. by a set of standard operating procedures. All biochemical analyses were performed in the same laboratory at SMHC, following standardized protocols. While there were inevitable changes in equipment over the 10-year period, we attempted to minimize its impact on our results. Reference ranges were standardized according to the laboratory’s protocols throughout the study period. The inpatients had neither tobacco use nor alcohol consumption at least 18 h before blood specimen collection. The exclusion criteria include: (1) Abnormal liver function caused by drugs such as valproate and agomelatine; (2) Significant hypothyroidism caused by lithium carbonate; (3) Abnormally increased blood lipids and blood glucose caused by the history of olanzapine; (4) Diabetes and coronary heart; (5) History of taking blood lipid-lowering drugs.

This study has been registered at International Clinical Trials Registry Platform (NCT03949218). This study was approved by the Institution Review Board of SMHC (IRB, number of 2019-15R), and the informed consent requirement was omitted according to relevant research. Patients’ personally identifiable information have been cleaned in order to protect their privacy.

Participants

1068 patients with bipolar disorder and 2015 patients with major depressive disorder were enrolled. 269 BD patients at mixed episodes (BDM), and 86 MDD patients with mixed characteristics (MDM) were selected. Basic information of enrolled patients including age, sex and disease status of patients (by the criteria of DSM-5 based on the information from their medical records). Biochemical indexes mainly consisted of five dimensions: (i) immune-inflammatory, including neutrophil% (NEUT%), erythrocyte sedimentation rate (ESR), white blood cell (WBC), C-reactive protein (CRP), albumin (ALB), globulin (GLO), albumin/globulin (A/G), prealbumin (PA); (ii) liver function, including total bilirubin (TBIL), direct bilirubin (DBIL), alanine aminotransferase (ALT), aspartate aminotransferase (AST), lactic dehydrogenase (LDH), gamma-glutamyl transferase (GGT), alkaline phosphatase (ALP); (iii) metabolism, including uric acid (UA), blood glucose (GLU), total cholesterol (CHOL), triglyceride (TG), high-density lipoprotein (HDL), low-density lipoprotein (LDL); (iv) thyroid hormones, including free triiodothyronine (FT3), free thyroxine (FT4), thyroid-stimulating hormone (TSH), total thyroxine (T4), total triiodothyronine (T3).

1155 patients at depressive episodes of major depressive disorder and 185 patients at depressive episodes of bipolar disorder were extracted to validate the distinguish model for BDM and MDM. Disease status of patients were defined by the criteria of DSM-5 based on the information from their medical records.

Statistical analysis

Statistical analysis was conducted with SPSS software (version 25.0) and R software (version 4.4.2; http://www.Rproject.org). The reported statistical significance levels were all two-sided, with statistical significance set at 0.05.

The normality of all continuous variables was assessed using the Shapiro-Wilk test along with evaluations of skewness and kurtosis. The characteristics features of BDM and MDM patients were compared by Student’s t-test for normally distributed data; otherwise, the Mann-Whitney U test was used; and Chi-square test for categorical variable. To identify any sex-based differences, comparisons of biochemical parameters between BDM and MDM patients were conducted by sex subgroups. Then univariate logistic regression was conducted to evaluate the association between the significant biochemical parameters above detected and disease status. Next, the multivariate logistic regression was performed to adjust for age and sex. The MDM group was set as the reference group in logistic regression analysis. Network analysis was employed to investigate the relationships among all the peripheral biomarkers in BDM and MDM groups separately using the R package ‘qgraph’ [36, 37]. For each group, pairwise correlations were computed between the clinical features using Pearson correlation coefficients. This resulted in a symmetric correlation matrix for each group, where the entries represent the strength of the association between the features. To estimate sparse and interpretable networks, the graphical least absolute shrinkage and selection operator (GLASSO) technique was applied for the correlation matrices. Besides, the Expected Influence was calculated as a measure of each node’s influence on its neighboring nodes. Then to compare the network structures and centrality measures between the BDM and MDM, the Network Structure Comparison and Centrality Comparison were employed using the R package ‘NetworkComparisonTest’. As a supplementary analysis, to further explore the interrelationships among the multiple biochemical markers and identify potential underlying patterns, we also conducted a Principal Component Analysis (PCA).

For post-hoc power analysis, we utilized G*Power software (version 3.1) to evaluate whether our existing sample sizes (269 BDM patients and 86 MDM patients) provided adequate statistical power for detecting meaningful differences between groups. Using a two-tailed independent samples t-test model with unequal sample sizes (n₁ = 269, n₂ = 86), alpha level (α) set at 0.05, and a medium effect size (d = 0.5), our post-hoc power analysis yielded a statistical power of 0.85 (85%). This exceeds the conventionally accepted minimum power threshold of 0.80 (80%), indicating that our sample size was sufficient for the primary between-group comparisons.

Results

Patients’ demographics and clinical characteristics

A total of 355 patients with mood disorders were included in the study, consisting of 269 BD patients at the mixed episodes, and 86 MDD patients with mixed characteristics. Table 1. reveals the differences of the general and clinical characteristics of patients between BDM and MDM. Immune-inflammatory related factors NEUT%, ESR and WBC showed higher levels in BDM, while CRP and PA were higher in MDM. Besides, MDM patients exhibited higher level of liver functional index DBIL, and lower level of LDH compared with BDM. Moreover, BDM patients had higher GLU level and lower peripheral TG level. For thyroid related hormones, FT3 and FT4 were higher in BDM subjects, while T4 was higher in MDM participants. Sex based subgroup comparison showed similar results (Supplementary Tables 1 and Table 2).

Table 1.

Comparison of clinical characteristics between MDM and BDM patients

Variables	MDM (n = 86)	BDM (n = 269)	χ²/z/t	p
Age	39.92(16.13)	34.57(14.31)	2.655	0.008**
Sex			0.868	0.352
Male	35(40.70%)	127(47.21%)
Female	51(59.30%)	142(52.79%)
NEUT%	56.65(11.52)	61.12(11.92)	-2.989	0.003**
ESR(mm/60min)	6.63(6.94)	8.12(6.95)	-2.368	0.018*
WBC(109/L)	6.64(1.87)	7.67(2.47)	-3.702	< 0.001**
CRP(mg/L)	3.65(4.81)	2.74(4.37)	2.254	0.024*
ALB(g/L)	41.38(3.95)	41.98(4.10)	-1.193	0.234
GLO(g/L)	27.47(3.65)	27.46(4.43)	0.486	0.627
A/G	1.53(0.24)	1.57(0.28)	-0.665	0.506
PA(mg/L)	304.53(87.38)	255.66(66.00)	4.886	< 0.001**
TBIL(µmol/L)	14.72(7.11)	14.58(6.77)	-0.202	0.840
DBIL(µmol/L)	4.05(2.04)	2.03(1.18)	8.729	< 0.001**
ALT(U/L)	26.24(23.57)	25.86(23.61)	-0.546	0.585
AST(U/L)	23.91(14.24)	23.20(13.11)	0.364	0.716
LDH(U/L)	134.37(28.54)	148.24(46.08)	-2.584	0.010*
GGT(U/L)	35.30(85.55)	19.99(15.36)	0.944	0.345
ALP(U/L)	65.75(22.29)	62.56(18.80)	1.342	0.180
UA(µmol/L)	316.69(72.49)	341.83(102.47)	-1.745	0.081
GLU(mmol/L)	5.15(1.03)	5.74(1.54)	-4.149	< 0.001**
CHOL (mmol/L)	4.70(1.06)	4.56(1.00)	0.944	0.345
TG(mmol/L)	1.48(0.96)	1.26(0.88)	2.995	0.003**
HDL(mmol/L)	1.36(0.32)	1.31(0.35)	1.514	0.130
LDL(mmol/L)	2.71(0.90)	2.55(0.86)	1.649	0.099
FT3(pmol/L)	4.17(1.15)	4.68(1.69)	-4.15	< 0.001**
FT4(pmol/L)	15.14(5.08)	16.97(6.16)	-2.417	0.016*
TSH(mU/L)	2.34(1.69)	2.77(2.12)	-1.517	0.129
T4(nmol/L)	96.18(29.07)	94.80(24.65)	2.065	0.039*
T3(nmol/L)	1.58(0.36)	1.62(0.44)	-0.827	0.408

Values are mean(SD) or number(percents%). p value is derived from the student’s t-test (normally distributed data) or the Mann-Whitney U test (non-normally distributed data). *p < 0.05; **p < 0.01

Table 2.

Univariate analysis and multivariate logistic regression of peripheral biochemicals

Variables	Univariate logistic regression			Multivariate logistic regression
	β	Odds ratio (95% CI)	p	β	Odds ratio (95% CI)	p
NEUT%	0.032	1.032(1.011–1.054)	0.003**	0.040	1.040(1.018–1.064)	< 0.001**
ESR	0.035	1.036(0.995–1.078)	0.086	0.079	1.083(1.030–1.138)	0.002**
WBC	0.216	1.241(1.097–1.405)	0.001**	0.206	1.229(1.085–1.392)	0.001**
CRP	-0.040	0.961(0.915–1.009)	0.111	-0.023	0.977(0.928–1.029)	0.382
PA	-0.009	0.991(0.987–0.994)	< 0.001**	-0.011	0.989(0.985–0.993)	< 0.001**
DBIL	-0.851	0.427(0.345–0.527)	< 0.001**	-0.886	0.412(0.331–0.513)	< 0.001**
LDH	0.010	1.010(1.003–1.018)	0.008**	0.012	1.012(1.004–1.021)	0.002**
GLU	0.426	1.531(1.187–1.974)	0.001**	0.538	1.712(1.307–2.242)	< 0.001**
TG	-0.235	0.790(0.616–1.014)	0.064	-0.188	0.828(0.639–1.074)	0.155
FT3	0.534	1.706(1.274–2.284)	< 0.001**	0.444	1.559(1.147–2.119)	0.005**
FT4	0.100	1.105(1.035–1.180)	0.003**	0.091	1.095(1.025–1.170)	0.007**
T4	-0.002	1.274(0.680–2.387)	0.663	-0.002	0.998(0.989–1.008)	0.752

Multivariate logistic regression was adjusted by age and sex

β is the regression coefficient. p value is derived from univariate and multivariate logistic regression. **p < 0.01; *p < 0.05

Associations between the peripheral biochemicals and disease status

Univariate and multivariate logistic regression analyses were performed to investigate the relationship between various peripheral biochemical markers and diagnosis status. In the univariate analysis, several biomarkers were significantly associated with the diagnosis outcome, including NEUT%, WBC, PA, DBIL, LDH, GLU, FT3 and FT4. For the multivariate logistic regression, age and sex were adjusted. The multivariate analysis confirmed that NEUT%, WBC, PA, DBIL, LDH, GLU, FT3 and FT4 remained significantly associated with the diagnosis outcome after controlling for these covariates. Among these factors, NEUT% (OR = 1.040, 95% CI:1.018–1.064, p < 0.001), WBC (OR = 1.229, 95% CI:1.085–1.392, p = 0.001), LDH (OR = 1.012, 95% CI:1.004–1.021, p = 0.002), GLU (OR = 1.712, 95% CI:1.307–2.242, p < 0.001), FT3 (OR = 1.559, 95% CI:1.147–2.119, p = 0.005) and FT4 (OR = 1.095, 95% CI:1.025–1.170, p = 0.007) showed positive association with the diagnosis of BDM, while PA (OR = 0.989, 95% CI:0.985–0.993, p < 0.001) and DBIL (OR = 0.412, 95% CI:0.333–0.510, p < 0.001) displayed negative correlation with BDM (Table 2).

Patterns of the peripheral biochemicals in MDM and BDM groups

The network analysis for the MDM group highlights the interrelationships between various variables, as depicted in the network plot and the expected influence plot (Fig. 1A). Key variables such as NEUT%, served as central hubs within the network. Highly influential variables include FT3, which is likely to have strong connections with multiple other variables. The BDM group also yielded a clear depiction of the interrelationships among the variables, as shown in Fig. 1B. Similarly, NEUT% appeared as hub node in the network as in the MDM group. FT4 ranks the highest in terms of the Expected Influence, and FT3 also exhibit high expected influence values. The Network Invariance Test showed no significant structural differences between the BDM and MDM networks (p = 0.422). The Global Strength Invariance Test indicates that, while the BDM network shows higher overall connectivity than the MDM network (Strength of BDM = 11.365, Strength of MDM = 4.752), this difference is not statistically significant (p = 0.238). Figure 2 displayed the results of PCA. The PCA of all biochemical markers revealed that the first two principal components explained 24.3% of the total variance (PC1: 13.7%, PC2: 10.6%). Analysis of variable loadings on PC1 showed that immune-inflammatory markers had the strongest contributions, with ESR, CRP, and WBC having prominent positive loadings. On PC2, thyroid hormones (particularly FT3 and FT4) and metabolic indicators (GLU) showed the strongest influence. This pattern aligns with our primary findings that immune-inflammatory factors and thyroid hormones exhibited significant differences between BDM and MDM patients.

Fig. 1.

Network of immune-inflammatory, liver function, metabolism, thyroid related factors and the expected influence plot for MDM (A) and BDM (B). In the network, each node represents a variable, and the edges connecting them represent the relationships between the variables. Blue edges indicate positive correlations, while red edges signify negative correlations. The thickness of the edges corresponds to the strength of the relationships, with thicker edges indicating stronger connections

Fig. 2.

Principal component analysis and variable loadings of biochemical indicators in MDM and BDM patients. (A) PCA score plot of all samples based on biochemical indicators. Each dot represents a patient, colored by diagnosis: MDM (red) and BDM (blue). The first two principal components (PC1 and PC2) explain 13.7% and 10.6% of the total variance, respectively. Ellipses indicate the 95% confidence interval for each group. (B) Variable loadings on PC1. Bars represent the contribution (loading value) of each biochemical variable to the first principal component. Positive and negative values reflect the direction and magnitude of the influence on PC1. (C) Variable loadings on PC2. Blue and red gradients indicate the strength and direction of the variable’s contribution

Discussion

The present study provides key insights into the biochemical differences between patients with bipolar disorder at mixed episodes and major depressive disorder with mixed features. Our findings reveal that BDM patients exhibit higher levels of immune-inflammatory markers such as NEUT% and WBC, along with increased thyroid hormones (FT3 and FT4), while MDM patients have elevated levels of CRP, DBIL, and PA. Logistic regression analysis showed that NEUT%, WBC, FT3, and FT4 were positively associated with the diagnosis of BDM, whereas PA and DBIL were negatively associated. Additionally, network analysis indicated no significant structural differences between the BDM and MDM groups, suggesting that while individual biochemical markers vary significantly, the overall network of peripheral biochemicals remains similar between the two conditions. Although the PCA did not show complete separation between two groups, the partial distinction observed in the distribution patterns provides additional support for the existence of biochemical differences between BDM and MDM conditions, particularly along dimensions related to inflammation and thyroid function.

There has been abundant of studies exploring the role of inflammation in mood disorders. Clinical evidence showed that patients with MDD exhibit increased level of CRP, which is an acute phase protein for systemic inflammation measurement [38]. CRP concentrations are also increased in BD and displayed higher levels during manic episodes [23]. It has been reported that higher level of CRP was noted in BD than in unipolar depression [24]. Interestingly, the current study showed adverse tendency, BD patients at mixed episodes exhibited lower level of CRP than MDD patients with mixed states, which suggested a more complicated inflammatory dysregulation in both disorders with mixed features. The higher CRP in MDM versus lower levels in BDM indicate that inflammation in mixed depressive states may resemble unipolar depression more closely, potentially driven by chronic stress or metabolic factors [39]. This pattern connects MDM to broader mood disorder pathophysiology, where CRP elevation is implicated in treatment-resistant depression and somatic comorbidities like cardiovascular disease [40], underscoring shared inflammatory pathways. Neutrophils represent the first line of immune defense, and NEUT% reflects the intensity of stress and systemic inflammation. A meta-analysis reported that subjects with MDD or BD had higher NEUT% as compared with health controls [41]. Comparisons of neutrophil between different bipolar phases or between BD and MDD are limited. A recent study found that NEUT% in BD patients at manic states was significantly higher than MDD that of patients, while BD patients at depressive episodes presented comparably elevated NEUT% [42]. Similar to previous evidence, NEUT% was higher in BDM individuals relative to MDM in this study. This inflammatory signature aligns with evidence of systemic immune dysregulation in mood disorders, where pro-inflammatory cytokines, linked to neutrophil activation, correlate with symptom severity and mood instability [43]. Prealbumin is a negative acute-phase reactant, which is downregulated by acute phase cytokines. Many studies have reported low PA levels in cerebro-spinal fluid (CSF) patients with MDD [44]. Frye MA et al. found that PA level was higher in BP-I and lower in BP-II compared with MDD [45]. This study provided further evidence that individuals with bipolar disorder at mixed episodes showed lower level of PA than MDD patients with mixed features.

There has been growing interests in the role of neuroendocrine hormones played in the mechanisms of mood disorders. The relationship between thyroid hormones, particularly FT3 and FT4, and mood disorders has been extensively studied [46, 47]. Previous evidence highlighted that the neuroendocrine abnormalities, including disruptions in thyroid hormone levels, are prevalent in untreated patients with both MDD and BD. This suggests that thyroid dysfunction might play a critical role in the underlying pathophysiology of these disorders, contributing to mood dysregulation and disease progression. Specifically, lower thyroid hormone secretion was found in MDD patients than in BD patients [46]. Song et al. reported significantly higher FT3 levels of BD patients at mania/hypomania episodes than individuals with BD at depressive episodes [48]. These findings imply that dysregulation of thyroid hormones may be associated with the onset and course of BD. Similarly, Zhao et al. also reported the level of FT3 in the manic group were significantly higher than those in the depressive group in BD samples [49]. This evidence points to the sensitivity of thyroid hormones to fluctuations in mood, further highlighting their role in the pathogenesis of mood disorders. In the current study, both FT3 and FT4 were higher in individuals with BDM compared to MDM. Based on previous evidence that (1) thyroid hormones were higher in BD patients than in MDD subject, and (2) in patients with BD diagnosis, those at manic episodes exhibited higher level of FT3 than those at depressive duration, we provided further information on the correlation of thyroid hormones and mixed status of mood disorders. Collectively, these findings underscore the significant role of thyroid hormones, particularly FT3 and FT4, in differentiating between BDM and MDM, and highlight their potential utility as biomarkers for monitoring disease course [50].

Beyond elucidating pathophysiological differences, these biochemical distinctions offer practical implications for clinical practice. First, the observed elevations in NEUT% and WBC in BDM patients, alongside higher CRP in MDM patients, suggest differential immune-inflammatory activation that could serve as an adjunctive diagnostic tool. Distinguishing BDM from MDM remains clinically challenging due to overlapping symptoms such as agitation, irritability, and mood lability. Incorporating NEUT% and WBC, readily measurable via routine complete blood counts, into diagnostic algorithms could provide an objective biomarker to differentiate BDM from MDM. For instance, a higher NEUT% or WBC might prompt clinicians to explore a bipolar spectrum diagnosis in patients initially presenting with depression, potentially reducing misdiagnosis rates and expediting appropriate management [39]. Second, these biochemical markers may have utility in monitoring disease progression and treatment response. The positive association of FT3 and FT4 with BDM aligns with prior evidence linking thyroid hyperactivity to manic or mixed states in bipolar disorder. Thyroid hormones are easily assessed in clinical settings and could be tracked longitudinally to evaluate shifts in mood states or treatment efficacy. For example, a decline in FT3 or FT4 levels in a BDM patient might indicate a transition from a mixed episode to a depressive state, signaling the need for therapeutic adjustment [51]. Integrating these markers into routine monitoring could thus provide clinicians with objective data to complement subjective symptom assessments, enhancing the precision of disease management. Third, these findings could inform targeted therapeutic strategies. The heightened immune-inflammatory profile in BDM (e.g., NEUT%, WBC) suggests that anti-inflammatory agents, such as adjunctive non-steroidal anti-inflammatory drugs (NSAIDs) or novel biologics, might be explored in clinical trials to mitigate mixed symptoms, particularly in patients with elevated inflammatory markers. This aligns with emerging evidence that inflammation modulates mood disorder severity [52]. These tailored approaches, grounded in biochemical profiles, could pave the way for personalized medicine in mood disorders with mixed features.

Limitations

Despite the valuable insights provided by this study, there are several limitations that need to be acknowledged. First, this is a retrospective study, which may introduce biases related to data collection and patient selection. Second, the smaller MDM sample size may limit the statistical power to detect significant associations within this subgroup, reducing the generalizability of our findings to broader MDM populations. To address this, future studies should employ larger, more balanced cohorts to validate our results and enhance the robustness and applicability of the observed biochemical differences between BDM and MDM. Third, while our exclusion criteria mitigated some medication-related effects, we did not fully adjust for the broader use of psychotropic medications, including other antipsychotics and antidepressants. These medications are known to influence some biochemical parameters. Similarly, lifestyle factors such as diet, BMI, smoking, and chronic alcohol consumption, which can alter inflammation and metabolism, were not systematically recorded. This lack of adjustment may confound our findings. Future studies should prospectively collect and control for these variables—through detailed patient histories and standardized questionnaire to strengthen the robustness of the observed associations and enhance their clinical relevance. Fourth, our approach is the exclusive focus on peripheral biochemical markers without corresponding assessments of clinical symptom severity, such as the Young Mania Rating Scale (YMRS) for manic symptoms or the Montgomery-Åsberg Depression Rating Scale (MADRS) for depressive symptoms. Combining these clinical severity scales with our biochemical data could significantly enhance diagnostic and prognostic precision. Lastly, the cross-sectional design precludes causal inferences, necessitating longitudinal research to clarify the relationships between these factors and biochemical profiles.

Conclusion

In conclusion, this study provides valuable insights into the peripheral biochemical distinctions between bipolar disorder at mixed episodes and major depressive disorder with mixed features. These differences, particularly in immune-inflammatory markers and thyroid hormones, emphasize the potential for biochemical markers to serve as diagnostic tools or indicators of disease progression. Future studies should focus on establishing causal relationships through prospective designs, integrating multi-modal approaches such as neuroimaging and genetic analyses, and expanding the scope to include larger, more diverse patient populations. Such efforts will not only help in elucidating the pathophysiology underlying mood disorders but also in developing targeted, individualized treatment strategies, ultimately improving clinical outcomes for individuals affected by mood disorders.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Xiaohui Wu and Shuo Wang the study’s design, data collection, and manuscript development. Zhiang Niu, Yuncheng Zhu, Ping Sun and Wenxi Sun supervised the initial analysis and contributed to manuscript drafting and revision. Yiru Fang and Jun Chen provided critical intellectual revisions and granted final approval for publication. All authors have approved the final manuscript and provided their consent for the public dissemination of the study.

Funding

The work was supported by the National Natural Science Foundation of China. (81930033), the Shanghai Mental Health Centre Clinical Research Center Project (CRC2018DSJ01-1, SHDC2020CR6023, CRC2021DX01).

Data availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethical approval and consent to participate

This study was approved by the Institution Review Board of SMHC (IRB, number of 2019-15R), and the informed consent requirement was omitted according to relevant research, which was approved by the ethics committee. Patients’ personally identifiable information have been cleaned to protect their privacy. Our study adhered to the Helsinki Declaration and applicable national guidelines.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.