Neurocognitive function profiles in Chinese patients with melancholic vs. non-melancholic major depressive disorder

Abstract

Background

The study aimed to investigate the neurocognitive characteristics of Chinese patients with melancholic depression (MEL) versus non-melancholic depression (NMEL).

Methods

Data were derived from a multicenter, multiphase prospective study conducted across nine top-tier hospitals in six provinces of China. A total of 287 healthy controls (HCs), 643 patients with major depressive disorder (MDD) were included. Melancholic traits were assessed using the Chinese version of the Mini-International Neuropsychiatric Interview (MINI). Cognitive function, encompassing processing speed, attention, verbal learning, visual learning, and executive function, was evaluated in both HCs and MDD patients.

Results

Among all MDD patients, 457 (71.07%) were classified as MEL. Both MEL and NMEL subgroups exhibited multi-domain cognitive impairment compared with HCs, including processing speed, attention, verbal learning, visual learning, and executive function (all P < 0.05). After Bonferroni correction, MEL patients showed significantly poorer performance on processing speed (Auditory Verbal Fluency, AVF) compared with NMEL patients (corrected P = 0.011). In contrast, no significant differences were observed between MEL and NMEL patients for executive function (Stroop Color-Word Test, SCWT, corrected P = 0.671) or any other cognitive domains (attention, verbal learning, visual learning). Multiple linear regression analysis confirmed that melancholic features were independently associated with poorer performance on both AVF test (B = -2.807, SE = 1.098, t = -2.556, P = 0.011, 95% CI = [-4.964, -0.651]) and SCWT (B = -2.360, SE = 1.039, t = -2.272, P = 0.023, 95% CI = [-4.400, -0.320]), after adjusting for potential confounders.

Conclusions

Both MEL and NMEL patients demonstrate multi-domain cognitive impairment, with more severe deficits in the MEL group. Melancholic features were a robust independent predictor of processing speed (AVF), whereas their association with executive function (SCWT) was only present in regression models and not significant after Bonferroni correction. These findings highlight the importance of cognitive assessment, especially for processing speed, in MEL patients.

Trial registration

ClinicalTrials.gov identifier: NCT02023567; registration date: 15 December 2013.

Background

For an extended period, there has been an ongoing debate in psychiatry whether melancholia constitutes a more severe form of depression or a distinct categorical entity [1–3]. The third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III) identifies melancholic features as one of the clinical subtypes of major depressive disorder, and the fifth edition (DSM-5) classifies it the same way [4, 5].

Over the past decade, extensive research has been conducted on the characteristic features of melancholia, though the findings from these studies have remained inconsistent. For example, some studies found that melancholia patients have more severe depressive symptoms [6–8], higher severity scores [9], and higher rates of unemployment [7], which was not confirmed in other studies [10].

In terms of cognitive function, some prior research has examined the differences between patients with melancholic depression (MEL) and those with non-melancholic depression (NMEL), however, the findings from these studies have been inconsistent. Some studies have reported distinct cognitive impairments in the MEL group. Marcos et al. (1994) and Withall et al. (2010) found that MEL patients exhibit significant cognitive deficits not only during the acute phase of illness but also after clinical recovery, suggesting that cognitive impairment in this subtype may be less responsive to symptom remission and could represent a more persistent trait [11, 12]. Valerio et al. (2021) conducted a systematic meta-analysis encompassing 14 independent studies and demonstrated that MEL patients perform significantly worse than NMEL patients across three key cognitive domains: attention, verbal fluency, and working memory [13]. This meta-analytic evidence strengthens the argument that MEL may be associated with more severe or domain-specific cognitive dysfunction compared to NMEL. Additionally, a prior study conducted in the Chinese population [14] revealed that, when compared with healthy controls, MEL patients displayed extensive cognitive deficits across most cognitive domains with the exception of attention, providing population-specific data that aligns with the broader trend of MEL-related cognitive impairment, though it did not directly compare MEL and NMEL groups.

In contrast, other studies have failed to detect meaningful differences in cognitive function between the two subtypes. For example, an early seminal study did not identify statistically significant differences in cognitive performance between MEL and NMEL patients, raising questions about the universality of cognitive distinctions between these subtypes [15]. More recently, Liang et al. (2023) focused on adolescent major depressive disorder (MDD) patients and assessed cognitive function using the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS), a widely validated neuropsychological tool [16]. Their results showed no significant differences in overall cognitive function or any of its subdomains between adolescent MDD patients with melancholic features and those with non-melancholic features, implying that age (specifically adolescent neurodevelopmental stages) may modulate the expression of cognitive differences between MEL and NMEL.

Given the inconsistencies in existing literature, compounded by small sample sizes in many prior studies and a lack of comprehensive comparisons across demographic and clinical variables, the current study aims to address these gaps by utilizing a large multicenter Chinese cohort, and systematically examining cognitive differences between MEL and NMEL patients while controlling for key confounders. We hypothesize that patients with MEL will exhibit more severe depressive symptoms and worse cognitive performance compared to those with NMEL.

Methods

Study participants and settings

Data for the present work were collected from the ODMPIM study (Objective Diagnostic Markers and Personalized Intervention in MDD Patients). The ODMPIM study is a multicenter, multiphase prospective investigation conducted across nine top-tier hospitals in six provinces of China. The study was approved by the Ethics Committee of Peking University Sixth Hospital (Approval No. 2013-29-1) and the relevant local ethics committees, and the present study was conducted in accordance with the ethical principles of the Declaration of Helsinki. All participants provided written informed consent prior to enrollment (ClinicalTrials.gov identifier: NCT02023567). The study design and procedural details have been fully described in previously published literature [17].

The inclusion criteria of the study were (1) an age between 18 and 55 years; (2) a diagnosis of MDD, either first-episode or recurrent, was confirmed using the Chinese version of the Mini-International Neuropsychiatric Interview (MINI) [18, 19]; (3) a total score on the 17-Item Hamilton Depression Rating Scale (HAMD-17) ≥ 14 [20]; (4) the ability to read and write to complete the questionnaire and psychological assessment; (5) no regular antidepressant use in the past two weeks or a change in antidepressants according to a psychiatrist’s advice; and (6) written confirmation of informed consent. Exclusion criteria for the study included the following: lifetime or current diagnosis of psychotic disorders or bipolar disorder, a history of alcohol or substance dependence, presence of severe physical illnesses, and being pregnant or breastfeeding. Healthy controls (HCs) with a current or lifetime history of mental disorders were excluded.

The detailed patient selection process is shown in Fig. 1. A total of 1,100 patients with MDD and 474 HCs were enrolled in the ODMPIM study cohort. For the current analysis, 930 participants who completed both standardized cognitive assessments and MDD subtype evaluations were selected, comprising 287 HCs and 643 MDD patients.

Fig. 1.

Patient selection procedures

Instrument and assessment procedure

Demographic Characteristics

A self-designed questionnaire was used to collect the following information from participants: age, age at onset of depression, gender, body mass index (BMI), years of education, duration of depressive illness, smoking status, employment status and family history of psychiatric disorders.

Depression diagnosis and clinical assessment

The HAMD-17 was administered to assess the severity of depressive symptoms. Consistent with an earlier study, MDD patients with melancholia were identified using the Chinese version of the MINI based on DSM-IV diagnostic criteria [21].

Cognitive function assessment

Trained investigators administered a cognitive test battery to evaluate five core cognitive domains: (1) information processing speed, evaluated via the Animal Verbal Fluency Scale (AVFS), Digit Symbol Coding Test (DSCT), Color Trial Test I (CTT-I), and the Stroop Color and Word Test (SCWT); (2) attention/vigilance, assessed via the Continuous Performance Test-Identical Pairs (CPT-IP); (3)verbal learning, assessed using the Hopkins Verbal Learning Test-Revised (HVLT-R); (4)visual learning, measured with the Brief Visual Memory Test-Revised (BVMT-R); and (5) executive function, evaluated through the SCWT and the Color Trial Test II (CTT-II)[17].All assessments were completed using standardized scales, except for the CPT, which was administered via a standardized computerized touchscreen.​ Raw scores of all cognitive tests were converted to T scores based on Chinese norms [22]. A higher score indicates better cognitive performance.

Statistical analysis

All statistical analyses were performed using IBM SPSS Statistics software (Version 27.0). Continuous variables were analyzed by one-way ANOVA (normal distribution) or the Kruskal–Wallis test (non-normal distribution). Categorical variables were analyzed using the chi-square test. Multivariate analysis of covariance (MANCOVA) was used to examine group differences in cognitive performance, with post hoc pairwise comparisons corrected via Bonferroni correction.

Within the overall MDD patient cohort, Pearson’s correlation analysis was conducted to explore the associations between demographic factors, clinical characteristics and the cognitive domains. Additionally, multiple linear regression analyses were conducted to determine whether melancholic features were independently associated with neurocognitive function (AVF test and SCWT scores) after adjusting for potential confounding variables, including age at onset of depression, gender, and years of education. Statistical significance was set at a two-tailed P < 0.05.

Results

Using the melancholic module of the Chinese version of the MINI, 457 (71.07%) patients with MDD were categorized as MEL. Demographic characteristics and clinical profiles of all participants are summarized in Table 1. Relative to HCs, both MEL and NMEL patients exhibited older age, lower educational attainment, a higher proportion of females, an increased prevalence of familial psychiatric history, and a greater likelihood of being unemployed (all P < 0.001). Compared with NMEL patients, MEL patients were older, had a later age at onset of depression, a shorter illness duration and a higher 17-item HAMD total score (all P < 0.05).​

Table 1.

Demographic and clinical characteristics of all subjects

	G1(n=287)	G2(n=186)	G3 (n=457)	χ2	P	Contrast
Female	158 (55.1%)	131 (70.4)	321 (70.2)	20.431	<0.001	G2, G3>G1
Smoker	48 (16.8%)	25 (13.5)	77 (16.9)	5.704	0.222
Employment status				60.515	<0.001	G1>G2, G3
Full-time job	256 (89.5)	132 (71.4)	300 (65.8)
Part-time job	9 (3.1)	4 (2.2)	11 (2.4)
No job	21 (7.3)	49 (26.5)	145 (31.8)
Family history of psychiatric disorders	6 (2.1)	32 (17.2)	76 (16.8)	39.918	<0.001	G2, G3>G1
Antidepressants	-	8 (4.3)	33 (7.2)	1.888	0.169
Current episode type	-			0.004	0.950
First episode	-	126 (68.5)	307 (68.2)
Recurrence	-	58 (31.5)	143 (31.8)
	Mean ± SD	Mean ± SD	Mean ± SD	t/Z	P	Contrast
Age (years)	34.3 ± 9.2	36.5 ± 10.5	39.2 ± 10.5	21.722	<0.001	G3>G2>G1
Age of onset (years)	-	32.7± 11.1	35.3 ± 11.1	-2.672	0.008	G3>G2
BMI	22.9 ± 3.2	22.7 ± 3.5	22.7 ± 3.5	0.281	0.755
Education (years)	14.8 ± 3.6	13.0 ± 3.7	12.6 ± 3.6	31.978	<0.001	G1>G2, G3
Duration (months, IQR)	-	6 (2, 12)	4 (2,12)	-2.066	0.039	G2>G3
HAMD-17 total score	-	19.2 ± 4.0	22.4 ± 5.2	-7.530	<0.001	G3>G2

G1=Healthy controls, G2=non-melancholic depression patients, G3=melancholic depression patients

Regarding cognitive function (Table 2), HCs outperformed both MEL and NMEL patients across multiple domains, including processing speed, attention, verbal learning, visual learning, and executive function (all P < 0.05). Compared with NMEL patients, MEL patients demonstrated worse performance in processing speed (assessed by the AVF, P = 0.001), and executive function (evaluated via the SCWT, marginally significant, P = 0.061). After Bonferroni correction, the difference in processing speed (AVF) remained statistically significant (corrected P = 0.011). For executive function (SCWT, corrected P = 0.671) and all other cognitive domains (attention, verbal learning, visual learning), no statistically significant differences were observed between MEL and NMEL patients.

Table 2.

Cognitive function of all subjects

	G1Mean ± SD	G2 Mean ± SD	G3 Mean ± SD	F	Pa	Pb	Contrastc
Speed of processing	46.3 ± 6.9	43.5 ± 7.9	42.1 ± 9.0	13.5	<0.001	0.484	G1>G2, G3
AVF	48.1 ± 11.1	45.9 ± 11.2	42.8 ± 12.3	8.579	0.002	0.011	G1, G2>G3
BASC	45.1 ± 14.0	41.6 ± 12.5	40.6 ± 13.1	6.332	0.008	1.177	G1>G2, G3
CTT-I	41.4 ± 10.4	38.5 ± 11.9	37.0 ± 11.5	16.294	<0.001	1.188	G1>G2, G3
Stroop-Word	47.9 ± 11.1	45.5 ± 11.2	44.8 ± 13.1	5.541	0.013	9.559	G1>G2, G3
Stroop-Color	48.8 ± 11.9	45.9 ± 10.7	45.1 ± 12.2	5.841	0.011	10.912	G1>G2, G3
Attention	44.6 ± 9.6	41.8 ± 10.2	42.1 ± 11.7	3.623	0.046	4.774	G1>G2, G3
Verbal learning	46.7 ± 10.9	42.5 ± 11.4	41.6 ± 11.3	10.906	<0.001	0.704	G1>G2, G3
HVLT-RTM-learning	45.7 ± 11.8	42.2 ± 12.1	40.9 ± 11.8	7.442	0.004	0.363	G1>G2, G3
HVLT-RTM-memory	47.8 ± 11.6	42.8 ± 12.5	42.3 ± 12.3	10.668	<0.001	2.167	G1>G2, G3
Visual learning	49.4 ± 10.8	45.7 ± 12.4	44.3 ± 13.0	9.139	0.001	2.585	G1>G2, G3
BVMT-RTM-learning	50.5 ± 13.0	47.2 ± 13.7	45.0 ± 14.0	6.660	0.006	1.672	G1>G2, G3
BVMT-RTM-memory	48.3 ± 10.4	44.2 ± 13.1	43.6 ± 13.6	9.063	0.001	5.093	G1>G2, G3
Executive function	47.4 ± 9.7	44.5 ± 9.2	42.6 ± 10.0	13.967	<0.001	0.374	G1>G2, G3
SCWT	51.7 ± 12.3	49.6 ± 10.3	47.0 ± 12.5	4.445	0.028	0.671	G1>G2, G3
CTT-II	43.2 ± 13.2	39.5 ± 12.5	38.1 ± 13.2	8.856	0.002	1.529	G1>G2, G3

G1=Healthy controls, G2=non-melancholic depression patients, G3=melancholic depression patients

^a P value was adjusted for age, gender, education, employment status, as study center as a random factor

^b Bonferroni-corrected P values for comparisons between non-melancholic and melancholic MDD

^c Bonferroni correction for post hoc pairwise comparisons

Group comparisons for these two measures (AVF and SCWT) between MEL and NMEL are presented in Fig. 2, the melancholic group displayed poorer performance on both than non-melancholic group (P < 0.05).

Fig. 2.

Comparison of non-melancholic and melancholic groups on AVF and SCWT

Correlational analyses (Table 3) revealed that a higher HAMD-17 total score was associated with worse performance on the AVF test (r = -0.084, P < 0.001) and SCWT performance (r = -0.095, P < 0.05). Additionally, older age, later age at onset of depression, female gender and no smoking were correlated with better performance on the SCWT (all P < 0.05). No significant relationship was observed between the duration of depressive episodes experienced by patients and their cognitive performance.

Table 3.

Correlations of demographic, clinical characteristics and cognitive performance of MDD patients

	Age	Age of onset	Gender	Education	HAMD-17 total score	Smoking	Duration
Speed of processing	0.178**	0.157**	0.249**	-0.009	-0.070	-0.064	0.042
AVF	0.024	0.010	-0.066	0.026	-0.084*	0.070	0.033
BASC	0.254**	0.226**	0.266**	0.014	0.013	-0.079*	0.044
CTT-1	0.070	0.093*	0.039	0.043	-0.155**	0.075	0.000
Stroop-Word	0.141**	0.106**	0.332**	-0.038	-0.034	-0.154**	0.064
Stroop-Color	0.137**	0.121**	0.381**	-0.077*	0.002	-0.129**	0.005
Attention	0.086*	0.088*	-0.022	-0.035	-0.029	0.017	-0.008
Verbal learning	0.061	0.063	0.143**	0.043	0.023	-0.002	0.049
HVLT-RTM-learning	0.085*	0.092*	0.132**	-0.012	0.017	-0.025	0.053
HVLT-RTM-memory	0.030	0.027	0.135**	0.090*	0.026	0.020	0.039
Visual learning	0.030	0.042	0.105**	0.060	0.038	-0.007	0.051
BVMT-RTM-learning	0.048	0.053	0.089*	0.030	0.039	-0.011	0.035
BVMT-RTM-memory	0.007	0.026	0.109**	0.083*	0.033	-0.002	0.061
Executive function	0.168**	0.174**	0.147**	-0.050	-0.126**	-0.009	-0.024
SCWT	0.142**	0.123**	0.296**	-0.063	-0.095*	-0.118**	-0.010
CTT-II	0.123**	0.150**	-0.050	-0.019	-0.104**	0.096*	-0.026

Multiple linear regression analysis (Table 4) showed that melancholic features remained independently associated with poorer performance on both the AVF (B = -2.807, SE = 1.098, t = -2.556, P = 0.011, 95% CI = [-4.964, -0.651]) and the SCWT (B = -2.360, SE = 1.039, t = -2.272, P = 0.023, 95% CI = [-4.400, -0.320]).

Table 4.

Multiple linear regression models of cognitive impairment and type of depressive episodes

Predictors	B	SE	β	t	p value	95%CI
Model: AVF, adjusted R2:0.017
Age of onset	0.046	0.046	0.042	0.991	0.322	(-0.045, 0.136)
Gender	-1.680	1.055	-0.064	-1.592	0.112	(-3.753, 0.392)
Education	0.046	0.139	0.014	0.328	0.743	(-0.227, 0.318)
Type of depressive episodes	-2.807	1.098	-0.106	-2.556	0.011	(-4.964, -0.651)
HAMD-17 total score	-0.170	0.099	-0.072	-1.724	0.085	(-0.364, 0.024)
Model: SCWT, adjusted R2:0.107
Age of onset	0.111	0.044	0.103	2.539	0.011	(0.025, 0.196)
Gender	7.506	1.107	0.288	6.781	<0.001	(5.332, 9.679)
Education	-0.091	0.132	-0.028	-0.688	0.491	(-0.349, 0.168)
Type of depressive episodes	-2.360	1.039	-0.090	-2.272	0.023	(-4.400, -0.320)
HAMD-17 total score	-0.220	0.093	-0.094	-2.363	0.018	(-0.402, -0.037)
Smoking	0.631	1.386	0.019	0.455	0.649	(-2.091, 3.353)

Type of depressive episodes: 0=non-melancholic depression, 1=melancholic depression; CI, confidence interval

Discussion

This is the first relatively large-scale, multicenter study investigating cognitive differences between MEL and NMEL patients in a Chinese depressed population. Using multiple linear regression analysis, we controlled for potential confounding factors that may influence cognitive performance, thereby improving the reliability and robustness of our results. Our findings demonstrated that melancholic features are independent predictors of processing speed (assessed via AVF) and executive function (assessed via SCWT).

In the present study, 71.07% of patients with MDD in our sample exhibited with melancholic features, a proportion falling within the previously reported range (16% to 76.2%) for prevalence of the melancholic subtype in MDD [6, 7, 21, 23–25]. Consistent with existing evidence, both the MEL and NMEL subgroups exhibited deficits in multiple neurocognitive domains (processing speed, attention, verbal learning, visual learning, executive function) compared with HCs. Impaired cognitive function during depressive episodes is well-documented in empirical research, with meta-analyses confirming multi-domain cognitive deficits as a core feature of depression compared to HCs [26–29]. Further aligning with prior work [13, 30–32], our findings validate that both MEL and NMEL patients exhibit multiple neurocognitive impairments compared to HCs.

Notably, MEL patients displayed more severe cognitive impairments than NMEL patients, with the most prominent deficits in processing speed. This observation is consistent with critical literature, including a large-scale outpatient study [31], a recent meta-analysis [13], and a Chinese cohort study [14], all of which identify processing speed impairments as a hallmark and robust cognitive feature of melancholia. This observation also echoes the clinical significance of psychomotor retardation, a core diagnostic feature of melancholia, and this core deficit is further supported by both Bonferroni-corrected group comparisons and independent multiple linear regression results in our study. Collectively, these results underscore that cognitive impairment, particularly in processing speed, is closely associated with melancholic features, highlighting its clinical significance as a key cognitive characteristic of MEL patients and providing valuable insights for the clinical assessment and differential diagnosis of MDD subtypes.

Beyond processing speed deficits, MEL patients showed a marginal trend toward poorer executive function (SCWT) compared to NMEL patients, though this difference did not reach statistical significance after Bonferroni correction for multiple comparisons. Importantly, multiple linear regression analysis–controlling for educational attainment [33], depression severity, and age at onset [34]–confirmed that melancholic features were independently associated with lower SCWT scores, which suggests a potential but not statistically robust association between melancholic features and executive function performance, albeit without significant group-level differences after correction. While prior literature [13, 35] has reported more severe executive function impairment in MEL patients relative to NMEL patients, our findings only support a tentative link between melancholic features and SCWT performance rather than a definitive subtype-specific deficit. The discrepancy between uncorrected group comparisons and regression results further emphasizes the necessity of controlling for demographic and clinical confounders when exploring subtype-specific cognitive differences in MDD.

Cognitive disturbances in MEL patients are hypothesized to relate to prefrontal-basal ganglia loops dysfunction, which modulate processing speed and executive function [36]. Neuroimaging studies have documented reduced activity and perfusion in the prefrontal and cingulate cortices, key brain regions mediating these cognitive domains [37, 38], supporting our observation of poorer cognitive performance in MEL patients and confirming that prefrontal cortical circuitry dysfunction is a hallmark of melancholia.

We also identified notable clinical and demographic differences between MEL and NMEL patients: MEL patients had a significantly higher total HAMD score, indicating greater depressive severity. Consistent with most prior studies [7–10, 39], MEL patients were older and had a later age at onset, which align with some studies [6], despite conflicting results in other studies [8, 10].

Methodologically, while categorical classification of MEL and NMEL based on DSM-IV criteria and the MINI is widely used, it has limitations owing to the clinical and biological heterogeneity of melancholia. Some studies have pointed out that DSM-based criteria lack precision, as they may reflected episode severity rather than qualitative differences between subtypes [35, 40]. As a post hoc analysis of the ODMPIM study, we did not employ nuanced assessment scales (e.g., CORE system [41], Newcastle Scale [42], or Sydney Melancholia Prototype Index (SMPI) [35]); Valerio et al. reported divergent cognitive findings when using SMPI versus DSM-5 criteria [35], underscoring the need for refined classification frameworks in future researches.

Building on the cognitive, clinical, and methodological insights elaborated above, key considerations pertaining to both result interpretation and clinical practice warrant attention. Regarding the potential impact of medication on baseline cognitive function, only 6.37% of our sample had prior antidepressant use; given that some antidepressants may affect cognitive function [43], this low proportion minimizes the risk of confounding effects on our cognitive outcomes. In terms of treatment implications, existing literature indicates that MEL subtypes exhibit differential treatment response [44], with pharmacotherapy potentially more efficacious than psychotherapy [45]. Accordingly, targeted interventions for MEL should incorporate their unique cognitive profiles and account for antidepressants’ cognitive impacts to optimize outcomes [43].

Limitations

Several limitations of the present study should be acknowledged. First, the cross-sectional design precludes establishing causality between melancholic features and cognitive function impairment, as well as the clarification of whether cognitive differences persist post-remission. Second, some confounding factors potentially associated with cognitive function, including hospitalization status [32] and comorbidity with psychotic symptoms [10, 34], were not assessed in this study. We further acknowledge that cumulative disease burden was not evaluated in the present study. Although current illness duration and episode type (first episode/recurrence) showed no significant correlation with cognitive performance, and no group difference in recurrence was observed, cumulative disease burden may have long-term neuroadaptive effects that were not captured by the current cross-sectional design. Future prospective studies should incorporate detailed assessments of cumulative burden to clarify its potential interaction with melancholic features in modulating cognitive outcomes. Third, the cognitive assessment battery did not cover all clinically relevant cognitive domains, such as working memory, decision-making, problem-solving, and planning [46]. Fourth, no neuroimaging data were collected, which prevents us from elucidating the underlying mechanisms driving the cognitive differences between the two groups. Fifth, the HCs and MDD groups are not matched, which may introduce potential biases. Future prospective longitudinal research incorporating key confounding variables, comprehensive assessments and mechanistic data is needed to address these limitations, clarify the causal relationship, and unravel the underlying mechanisms linking melancholic features to cognitive impairment. Despite these limitations, the large-scale Chinese multicenter sample ensures that our results reflect the actual cognitive function of MEL and NMEL patients in real-world clinical settings.

Conclusions

In conclusion, both MEL and NMEL patients exhibit multi-domain cognitive deficits relative to HCs, with MEL patients presenting more severe impairments overall. Crucially, melancholic features were a statistically robust independent predictor of poorer processing speed (AVF), a finding supported by Bonferroni-corrected group comparisons and multiple linear regression. Melancholic features were also independently associated with executive function (SCWT) in regression models, though no significant group differences were observed after Bonferroni correction. These findings validate cognitive heterogeneity among MDD subtypes and highlight processing speed impairment as a key clinical marker for melancholic MDD, thus informing differential diagnosis and individualized intervention.

Abbreviations

MEL

Melancholic depression

NMEL

Non-melancholic depression

HCs

Healthy controls

MDD

Major depressive disorder

MINI

Mini-International Neuropsychiatric Interview

AVF

Auditory Verbal Fluency test

SCWT

Stroop Color-Word Test

DSM-III

Diagnostic and Statistical Manual of Mental Disorders

RBANS

Repeatable Battery for the Assessment of Neuropsychological Status

ODMPIM

Objective Diagnostic Markers and Personalized Intervention in MDD Patients

HAMD-17

17-Item Hamilton Depression Rating Scale

BMI

Body mass index

DSCT

Digit Symbol Coding Test

CTT

Color Trial Test

CPT-IP

Continuous Performance Test-Identical Pairs

HVLT-R

Hopkins Verbal Learning Test-Revised

BVMT-R

Brief Visual Memory Test-Revised

SMPI

Sydney Melancholia Prototype Index

Author contributions

XY, TS, YS, LX conceived and coordinated the study, designed, and performed the experiments, analyzed the data, and wrote drafted the manuscript. LX, JL, HF, FY, and TS collected and analyzed data, and revised the manuscript. All authors read and approved the final version of the manuscript.

Funding

This work was supported by the National Key Basic Research Program of China (973 Program) (NO. 2013CB531305), the National Natural Science Foundation of China (NO. 82371530) and the National Key Research and Development Program of China (NO. 2025YFA1309200).

Data availability

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

Declarations

Ethics approval and consent to participate

All eligible participants provided written informed consent prior to enrollment in accordance with the ethical principles of the Declaration of Helsinki. The study protocol was reviewed and approved by the research ethics committee of each participating institution, namely Peking University Sixth Hospital, Peking University Institute of Mental Health; Nankai University Affiliated Anding Hospital (Tianjin Mental Health Center); Peking Union Medical College Hospital; First Hospital of Shanxi Medical University; The First Affiliated Hospital of China Medical University; Dalian Seventh People’s Hospital; Beijing Anding Hospital, Capital Medical University; The First Hospital of Hebei Medical University; and Tianjin Medical University General Hospital.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.