Correlation between Aβ_1–42, Dnmt3a2, urinary AD7c-NTP and cognitive dysfunction in first-episode and recurrent MDD: A case–control study

Abstract

Background and Aim:

Major depressive disorder (MDD) is one of the most prevalent mental illnesses worldwide and involves cognitive dysfunction that may negatively impact clinical and social outcomes. Previous studies have suggested that beta-amyloid peptide (Aβ_1–42), DNA methyltransferase (Dnmt3a2), and urinary Alzheimer-associated neuronal thread protein (AD7c-NTP) are associated with cognitive impairment. However, there are no relevant studies in MDD. The aim of this study was to assess the correlation between serum Aβ_1–42, Dnmt3a2, and urinary AD7c-NTP and cognitive dysfunction in MDD.

Materials and Methods:

A total of 59 eligible patients were included in the study, including 29 patients with first-episode MDD (FEDs) and 30 patients with recurrent MDD (RMDDs), and 30 matched healthy controls (HCs) were selected. Participants’ cognitive functioning was evaluated using the MATRICS consensus cognitive battery (MCCB). The enzyme-linked immunosorbent assay (ELISA) method was used to measure the concentrations of the three proteins. Statistical analysis was completed using Statistical Package for the Social Sciences (SPSS) 20.0. The statistical significance was set as P < 0.05.

Results:

Serum Dnmt3a2 and urinary AD7c-NTP showed significant differences among the three groups (both P < 0.001), but there were no significant differences in Aβ_1–42 levels. Upon examining the results of cognitive testing, we found that serum Aβ_1–42 was negatively associated with working memory scores in RMDDs (P = 0.020), but Dnmt3a2 was positively associated with working memory and verbal learning scores in the same cohort (P = 0.012 and P = 0.037, respectively). In contrast, urinary AD7c-NTP was negatively correlated with verbal learning scores in FEDs (P = 0.013).

Conclusions:

Serum Dnmt3a2 and Aβ_1–42 levels may be associated with cognitive impairment in RMDDs and may act as potential biomarkers of cognitive impairment. Although urinary AD7c-NTP was closely related to cognitive dysfunction in FEDs, this relationship did not hold in RMDDs.

INTRODUCTION

Major depressive disorder (MDD) is one of the most prevalent mental illnesses in the general population and is a major cause of disability worldwide.[1] It is well established that cognitive dysfunction (mainly involving deficits in executive functioning, sustained attention, processing speed, visual and verbal learning, and memory) is a core dimension of MDD.[2] Replicated evidence indicates that cognitive deficits can cause psychosocial impairment in severe mental disorders and influence patients’ social relationships, occupational performance, and overall quality of life.[1,3,4]

Previously, depressive cognitive impairment was thought to be transient. However, recent clinical observations and researches suggest that it may function as an antecedent to the formal diagnosis of MDD and can persist during “asymptomatic” states.[5,6] In addition, there are varying degrees of cognitive dysfunction in MDD patients, no matter whether they are in remission or in an acute disease period,[7,8] and this dysfunction may be one of the major risk factors for the recurrence of MDD. Additionally, cognitive impairment in recurrent MDD (RMDD) cases may be more severe than that in first-episode MDD (FED),[6] a finding which also suggests that clinicians should pay attention to the treatment of residual symptoms.

The biological mechanisms of cognitive dysfunction in MDD, however, have not been well elucidated. Macroscopically, ample evidence suggests that MDD is associated with atrophy in the hippocampus, frontal cortex, amygdala, ventral striatum, and atrophy in other brain areas,[9,10] all areas that are highly involved in cognitive function. Microscopically, neuroinflammation has been increasingly linked to adverse memory outcomes. In particular, the relationship between cholinergic circuit dysfunction and cognitive decline associated with neurodegenerative disorders has become an area of increasing study.[11,12] The fact that neural stem cells in the hippocampus express acetylcholine receptors (AChR), including nicotinic (nAChR) and muscarinic (mAChR), provides a potential mechanism by which the cholinergic system could influence neurogenesis.[12,13] Additionally, the alpha7 nicotinic acetylcholine receptor (α7 nAChR), a specific nAChR subtype, is known to be functionally important in the cholinergic system that modulates hippocampal memory processes.[12-14]

Several previous studies have shown that α7 nAchR is closely related to the neurotoxic actions of beta-amyloid peptide (Aβ_1–42).[14,15] Under normal physiological conditions, the production and degradation of the Aβ peptide are in a dynamic balance in the human central nervous system. There are two main forms of Aβ: Aβ_1-40 and Aβ_1–42. Aβ_1–42 has very few components, but it is neurotoxic and tends to aggregate rapidly. These changes may inhibit α7 nAChR-dependent neurotransmission through high-affinity binding and direct interaction with α7 nAChR, leading, in turn, to synaptic loss, neuronal death, and cognitive deficits.[16-18] This process can also activate the neuroimmune system, induce the secretion of proinflammatory cytokines in immune cells within and outside of the nervous system,[19,20] and significantly reduce the expression levels of neurotrophic factors and proteins,[21] which can eventually lead to emotional and cognitive impairment.[22,23]

Alzheimer-associated neuronal thread protein (AD7c-NTP) is a neuronal axon transmembrane protein that is thought to be involved in neurodegeneration-related neuritic sprouting and cell death.[24,25] Previous studies have suggested that cerebrospinal fluid (CSF) and urinary levels of AD7c-NTP have high sensitivity and specificity[26] and that AD7c-NTP may be an important biomarker of mild cognitive impairment, even early in the disease course.[27,28] Further research found that AD7c-NTP is a specific indicator that can predict Aβ deposition in patients with Alzheimer’s disease (AD) and mild cognitive impairment.[29] However, few studies have shown their relationship with cognitive dysfunction in MDD.

DNA methyltransferase (Dnmt3a2) is another key protein with neuroprotective effects that is involved in the memory process. Numerous studies have demonstrated that decreases in Dnmt3a2 levels are closely related to cognitive impairment,[30,31] and animal studies have found that Dnmt3a2 overexpression in the hippocampus of young adult mice induces memory enhancement in a variety of situations.[32,33]

Thus, this study aimed to explore the correlation between serum Aβ_1–42, Dnmt3a2, and urinary AD7c-NTP and cognitive dysfunction in patients with MDD. Also, our aim was to provide objective clinical biomarkers for evaluating cognitive dysfunction in RMDD through exploring some characteristic neurotoxic or neuroprotective proteins.

MATERIALS AND METHODS

Participants

The case group included 29 FED patients and 30 RMDD patients. They were recruited from treatment-seeking outpatients/inpatients of the First Affiliated Hospital of Zhengzhou University (China) between November 2016 and December 2017, using the diagnostic criteria for MDD in the Diagnostic and Statistical Manual of Mental Disorders IV (DSM-IV) and the 24-item Hamilton Depression Rating Scale (HAM-D-24) scores ≥21. Other inclusion criteria included patients who had never been treated with drugs or had previously taken antidepressants but had relapsed after a period of stopping medication (to minimize drug interference). Exclusion criteria were presence of serious physical or neurological comorbidities, pregnant or lactating women, alcohol dependence or psychoactive substance abuse, bipolar disorder, schizophrenia, other psychotic disorders, and organic mental disorders.

The control group included 30 subjects. Inclusion criteria included no known physical or mental disorders, no drug or alcohol abuse, no family history of psychosis, and HAM-D-24 scores <7.

All the participants needed to have at least attended primary school (in order to understand the content of the assessment), and their age range was from 18 to 60 years. All protocols were approved by the Ethics Committee of Zhengzhou University, and written informed consent was obtained from all the participants.

Cognitive function assessments

Cognitive functioning was evaluated in a single session without break using the MATRICS consensus cognitive battery (MCCB) when patients gave informed consent to be enrolled. The following aspects are included in the Chinese version:

• (1)Trail making test (TMT), symbol coding (SC), and category fluency (CF), adopted to evaluate the speed of processing (SOP);
• (2)Working memory (Wechsler Memory Scale-III [WMS-III], digital span, and spatial span);
• (3)Attention/vigilance (continuous performance test-identical pairs [CPT-IP]), verbal learning (Hopkins verbal learning test-revised [HVLT-R]);
• (4)Visual learning (brief visual memory test-revised [BVMT-R]);
• (5)Reasoning and problem solving (neuropsychological assessment battery [NAB], mazes); and
• (6)Social cognition (Mayer–Salovey–Caruso emotional intelligence test [MSCEIT], managing emotions).

Raw scores were entered into the professional software, which transformed them into seven domains of cognitive functioning.

Sample collection

Serum collection: Five milliliters of fasting cubital venous blood was collected between 7:00 and 8:00 am the day after enrollment (to minimize circadian variability) and saved in an ethylenediaminetetraacetic acid (EDTA) anticoagulant tube. The serum was centrifuged for 10 min (3000 rpm at 4°C). The upper layer of the serum was separated, uniformly numbered, and stored at −70°C for protein determination.

Urine collection: Fifteen milliliters of urine was collected between 7:00 and 8:00 am the day after study enrollment and stored for 20 min in the dark. The urine was centrifuged for 10 min (3000 rpm). The upper layer of the urine was separated, uniformly numbered, and stored at −70°C for protein determination.

Sample processing

Serum Aβ_1–42, Dnmt3a2, and urinary AD7c-NTP concentrations were assayed using an enzyme-linked immunosorbent assay (ELISA) kit (Anqun Biological Technology Co. Ltd., Shenzhen, China) according to the manufacturer’s instructions. Briefly, we completed the following steps:

• (1)One hundred microliters of the urine specimen or serum sample was placed into the precoated plate and incubated at 37°C for 1 h.
• (2)The plate was loaded with 100 μl of biotinylated rabbit anti-AD7c-NTP, anti-Aβ_1–42, or anti-Dnmt3a2 antibodies and incubated at 37°C for 1 h.
• (3)The plate was loaded with 100 μl of horseradish peroxidase-labeled avidin and incubated at 37°C for 30 min. Each of the above steps was followed by washing with phosphate-buffered saline (PBS) 5 times.
• (4)Fifty microliters of chromogenic reagents A and B were added in turn, and the mixture was incubated at 37°C for 15 min.
• (5)The reaction was stopped by adding 50 μl of sulfuric acid as a stop buffer.

The optical density (OD) value of the reaction product was measured at 450 nm (630 nm as a reference) with a Varioskan^® Flash Spectral Scanning Multimode Reader (Thermo Fisher Scientific, Waltham, MA, USA) within 15 min of stopping the reaction. The Aβ_1–42, AD7c-NTP, and Dnmt3a2 concentrations were positively correlated with the OD values and then calculated according to the standard curve of recombinant human AD7c-NTP, Aβ_1–42, and Dnmt3a2 peptides.

Statistical analysis

We used Statistical Package for the Social Sciences (SPSS) statistical software (version 20.0) for all statistical analyses. Data were expressed as mean ± standard deviation for continuous variables and frequencies and percentages for categorical variables. The Shapiro–Wilk (S–W) test was used to determine whether the data obeyed a pattern of normal distribution. Also, comparative analyses were done using t-tests, one-way analysis of variance (ANOVA), and Chi-square tests, respectively. The Student–Newman–Keuls test (SNK-t) was used for further post hoc examination, and the Scheffe procedure method was used for correction. The correlation analyses were done using Pearson’s correlation test. The statistical significance (two-tailed) was set as P < 0.05.

RESULTS

Comparison of demographic and clinical characteristics between cases and controls

The demographic data and brief clinical characteristics of the 89 study subjects are listed in Table 1. There were no statistically significant differences in gender, age, and educational level among the three groups (P > 0.05). There was no significant difference in HAM-D-24 scores between FEDs and RMDDs (P > 0.05) [Table 1].

Table 1.

Comparison of demographic and clinical characteristics between cases and controls

Group	FEDs (n=29)	RMDDs (n=30)	HCs (n=30)	χ2/t/F	P
Gender
Male, n (%)	14 (48.3)	9 (30.0)	15 (50.0)	3.0	0.22
Female, n (%)	15 (51.7)	21 (70.0)	15 (50.0)
Age, years	31.1±6.2	34.9±9.7	33.1±7.8	1.6	0.20
Educational level
High school or less, n (%)	12 (41.4)	15 (50.0)	11 (36.7)	1.1	0.57
At least some college or more, n (%)	17 (58.6)	15 (50.0)	19 (63.3)
HAM-D-24 scores	36.1±5.4	35.7±6.4	-	0.3	0.78
Duration of total illness, months	5.4±2.2	32.1±11.8	-	-12.2	<0.001
Numbers of episodes, n	-	3.4±1.8	-	-	-

FEDs=Patients with first-episode depression, HAM-D-24=24-item Hamilton Depression Rating Scale, HCs=Healthy controls, RMDDs=Patients with recurrent depression

Comparison of three protein levels between cases and controls

The serum concentration of Aβ_1–42 showed no significant differences among the three groups (F = 2.6, P = 0.077), but was higher in RMDDs than in healthy controls (HCs; P ＜ 0.05). In contrast, the serum Dnmt3a2 levels were lower in RMDDs and FEDs than in HCs (F = 10.8, P < 0.001), but showed no significant differences between RMDD and FED patients (P > 0.05). Urinary AD7c-NTP levels were higher in RMDDs and FEDs than in HCs (F = 47.1, P < 0.001) [Table 2].

Table 2.

Comparison of three protein levels between cases and controls

Group	FEDs (n=29)	RMDDs (n=30)	HCs (n=30)	F	P
Aβ1-42 (ng/ml)	9.7±11.8	15.7±21.4	6.4±12.3	2.6	0.077
Dnmt3a2 (U/L)	0.17±0.03	0.19±0.04	0.23±0.07	10.8	<0.001*
AD7c-NTP (pg/ml)	986.1±96.2	1028.9±102.1	817.4±64.9	47.1	<0.001*

FEDs=Patients with first-episode depression, HAM-D-24=24-item Hamilton Depression Rating Scale, HCs=Healthy controls, RMDDs=Patients with recurrent depression. *P<0.001

Cognitive function differences between cases and controls

Upon examining the results of cognitive testing, we found that working memory scores showed differences among all three groups (i.e., FEDs, RMDDs, and HCs) (F = 14.1, P < 0.001). These test scores were lower in RMDDs than both FEDs and healthy participants (P < 0.05), but there was no significant difference between FEDs and HCs (P > 0.05) [Table 3].

Table 3.

Cognitive function differences between cases and controls

Cognitive function	FEDs (n=29)	RMDDs (n=30)	HCs (n=30)	F	P
SOP	42.4±8.4	36.9±9.9	39.3±7.5	2.6	0.08
WMS-III	42.7±6.6	31.3±10.0	47.6±8.6	14.1	<0.001*
CPT-IP	40.7±7.4	40.8±6.4	40.5±7.4	0.2	0.79
HVLT-R	40.1±7.6	40.2±10.8	34.4±4.6	1.5	0.22
BVMT-R	41.0±12.5	36.4±9.8	40.7±8.6	0.8	0.46
NAB	38.4±10.0	39.0±8.8	41.4±9.2	0.4	0.67
MSCEIT	35.2±12.6	33.2±6.3	36.2±8.8	0.7	0.51

BVMT-R=Brief visual memory test-revised (visual learning), CPT-IP=Continuous performance test-identical pairs (attention/vigilance), FEDs=Patients with first-episode depression, HCs=Healthy controls, HVLT-R=Hopkins verbal learning test-revised (verbal learning), MSCEIT=Mayer-Salovey-Caruso emotional intelligence test (social cognition), NAB=Neuropsychological assessment battery (reasoning and problem solving), RMDDs=Patients with recurrent depression, SOP=Speed of processing, WMS-III=Wechsler Memory Scale-III (working memory). *P<0.001

Correlation analysis between the three protein levels and cognitive test scores

We observed that the serum concentration of Aβ_1–42 was negatively associated with working memory scores in RMDDs (r = −0.481, P = 0.020), whereas the serum Dnmt3a2 concentration was positively correlated with both working memory scores and verbal learning scores (r = 0.514, P = 0.012 and r = 0.382, P = 0.037 for working memory and verbal learning, respectively) in RMDDs [Table 4a]. In contrast, the urinary concentration of AD7c-NTP was negatively correlated with the verbal learning scores in FEDs (r = −0.456, P = 0.013) [Table 4b]. However, serum Aβ_1–42 and Dnmt3a2 levels were not correlated with cognitive test scores in FEDs or HCs (P > 0.05) [Tables 4b and 4c].

Table 4a.

Pearson’s correlation values for the relationship between protein levels and cognitive test scores in RMDDs

	SOP (r, P)	WMS-III (r, P)	CPT-IP (r, P)	HVLT-R (r, P)	BVMT-R (r, P)	NAB (r, P)	MSCEIT (r, P)
Aβ1-42 (ng/ml)	-0.080, 0.674	-0.481, 0.020*	-0.171, 0.367	-0.102, 0.593	-0.113, 0.553	-0.055, 0.772	0.117, 0.538
Dnmt3a2 (U/L)	0.308, 0.098	0.514, 0.012*	0.268, 0.151	0.382, 0.037*	0.336, 0.070	0.185, 0.328	0.013, 0.946
AD7c-NTP (pg/ml)	-0.028, 0.885	-0.405, 0.055	0.041, 0.830	0.077, 0.684	-0.099, 0.604	0.056, 0.769	-0.328, 0.077

Aβ_1-42=Beta-amyloid peptide, AD7c-NTP=Urinary Alzheimer-associated neuronal thread protein, BVMT-R=Brief visual memory test-revised (visual learning), CPT-IP=Continuous performance test-identical pairs (attention/vigilance), Dnmt3a2=DNA methyltransferase, HVLT-R=Hopkins verbal learning test-revised (verbal learning), MSCEIT=Mayer-Salovey-Caruso emotional intelligence test (social cognition), NAB=Neuropsychological assessment battery (reasoning and problem solving), RMDDs=Patients with recurrent depression, SOP=Speed of processing, WMS-III=Wechsler Memory Scale-III (working memory). *P<0.05

Table 4b.

Pearson’s correlation values between protein levels and cognitive test scores in FEDs

	SOP (r, P)	WMS-III (r, P)	CPT-IP (r, P)	HVLT-R (r, P)	BVMT-R (r, P)	NAB (r, P)	MSCEIT (r, P)
Aβ1-42 (ng/ml)	0.102, 0.599	-0.112, 0.562	0.218, 0.256	0.016, 0.935	0.150, 0.438	-0.177, 0.359	0.012, 0.955
Dnmt3a2 (U/L)	0.099, 0.610	0.020, 0.918	-0.008, 0.967	-0.142, 0.461	-0.127, 0.513	-0.134, 0.487	0.127, 0.535
AD7c-NTP (pg/ml)	-0.142, 0.464	-0.021, 0.912	-0.361, 0.055	-0.456, 0.013*	-0.301, 0.112	0.109, 0.575	-0.122, 0.551

Aβ_1-42=Beta-amyloid peptide, AD7c-NTP=Urinary Alzheimer-associated neuronal thread protein, BVMT-R=Brief visual memory test-revised (visual learning), CPT-IP=Continuous performance test-identical pairs (attention/vigilance), Dnmt3a2=DNA methyltransferase, FEDs=Patients with first-episode depression, HVLT-R=Hopkins verbal learning test-revised (verbal learning), MSCEIT=Mayer-Salovey-Caruso emotional intelligence test (social cognition), NAB=Neuropsychological assessment battery (reasoning and problem solving), SOP=Speed of processing, WMS-III=Wechsler Memory Scale-III (working memory). *P<0.05

Table 4c.

Pearson’s correlation values between protein levels and cognitive test scores in HCs

	SOP (r, P)	WMS-III (r, P)	CPT-IP (r, P)	HVLT-R (r, P)	BVMT-R (r, P)	NAB (r, P)	MSCEIT (r, P)
Aβ1-42 (ng/ml)	-0.156, 0.627	0.336, 0.286	0.279, 0.379	0.402, 0.196	0.145, 0.652	0.492, 0.104	0.155, 0.670
Dnmt3a2 (U/L)	-0.331, 0.270	0.055, 0.859	-0.094, 0.760	-0.028, 0.928	0.451, 0.122	-0.082 0.790	-0.028, 0.936
AD7c-NTP (pg/ml)	-0.373, 0.209	0.082, 0.790	-0.173, 0.571	0.186, 0.544	-0.422, 0.151	-0.199 0.514	-0.183, 0.591

Aβ_1-42=Beta-amyloid peptide, AD7c-NTP=Urinary Alzheimer-associated neuronal thread protein, BVMT-R=Brief visual memory test-revised (visual learning), CPT-IP=Continuous performance test-identical pairs (attention/vigilance), Dnmt3a2=DNA methyltransferase, HCs=Healthy controls, HVLT-R=Hopkins verbal learning test-revised (verbal learning), MSCEIT=Mayer-Salovey-Caruso emotional intelligence test (social cognition), NAB=Neuropsychological assessment battery (reasoning and problem solving), SOP=Speed of processing, WMS-III=Wechsler Memory Scale-III (working memory). *P<0.05

DISCUSSION

The current study was designed to explore the correlation between serum Aβ_1–42, Dnmt3a2, and urinary AD7c-NTP and cognitive impairment in MDD patients. Here, we observed that there were no significant differences in serum Aβ_1–42 levels among the three groups. There was no significant association between Aβ_1–42 levels and cognitive functioning in patients with FED, but we found that serum Aβ_1–42 levels were negatively correlated with working memory scores in patients with RMDD. However, a previous population-based cross-sectional study suggested that Aβ_1–42 levels and the Aβ_1–42/Aβ_1-40 ratio were elevated in AD patients with possible cognitive impairment and might be more pronounced in the early phases of cognitive impairment.[34] In addition, Aβ_1–42 abnormalities have been reported in late-life depression in previous studies.[35,36] A recent systematic review demonstrates that late-onset depression appears to be indicative of early neurodegeneration and stems from a vast array of cerebral abnormalities including white matter hyperintensities and Aβ and prefrontal dysfunction.[37] Our study further reveals that the association between cognitive impairment and Aβ_1–42 can also be observed in young patients with RMDD, which can be explained by the duration of disease or the number of episodes.

Another important result from the present study was that serum Dnmt3a2 levels were decreased in MDD patients, and that these levels were significantly positively correlated with working memory and verbal learning scores in RMDD patients. Animal studies have demonstrated that Dnmt3a2 positively regulates the expression of plasticity-related genes in cultured hippocampal neurons,[33] and that it facilitates cognition in the adult mouse hippocampus. These findings suggest that Dnmt3a2 is at the core of memory processes and hint that it could be a novel target for cognition-enhancing therapies, which could ameliorate anxiety and fear disorders and boost memory consolidation.[32] Thus, Dnmt3a2 may be a candidate biomarker to predict cognitive decline or relapse in depression. To the best of our knowledge, no previous studies have reported serum Dnmt3a2 levels in MDD patients.

In addition, we found that urinary AD7c-NTP levels in MDD patients were high compared to controls. However, contrary to our expectations, only FED patients had negative associations with verbal learning scores. In fact, the correlation between urinary AD7c-NTP levels and the severity of cognitive dysfunction remains controversial. Previous studies have shown that urinary AD7c-NTP levels are positively correlated with AD severity[25] and are significantly higher in cases of advanced AD than they are in early-stage disease.[38] Nevertheless, some studies suggest that it is not clear whether urinary AD7c-NTP can be used as an early marker,[39] and others have failed to find any relationships between AD7c-NTP levels and severity of cognitive function at all.[29] It is worth noting, however, that one controlled pilot study suggested that urinary AD7c-NTP levels appeared to be associated with cognitive impairment and could potentially be biomarkers for early identification of cognitive dysfunction in late-life depression.[40] However, another recent case–control study showed that, although urinary AD7c-NTP may be elevated in cases of severe cognitive decline, it still might not be a biomarker for subjective cognitive decline and/or depressive status.[41]

Limitations of the study

The study findings may be interpreted in light of certain limitations. First, the patients with FED in our study might not have exhibited sufficiently severe cognitive dysfunction to influence changes in serum Dnmt3a2 and Aβ_1–42 levels. Second, the sample size was too small to detect meaningful differences between MDD patients and HCs. Third, as a cross-sectional study, any long-term observation of the dynamic relationship between serum Aβ_1–42, Dnmt3a2, urinary AD7c-NTP levels and cognitive dysfunction was limited. Fourth, the cognitive impairments in MDD may have different underlying mechanisms than those in AD. The prognostic effect of urinary AD7c-NTP on cognitive dysfunction with MDD requires further study. Thus, it will be important to continue this line of research, expand the sample sizes, and carry out longitudinal studies in order to further confirm the relationships between serum Aβ_1–42, Dnmt3a2, urinary AD7c-NTP and cognitive dysfunction in MDD patients.

CONCLUSIONS

In conclusion, serum Dnmt3a2 and Aβ_1–42 levels may be closely correlated with cognitive dysfunction in patients with RMDD and may act as potential biomarkers of cognitive impairment in depression. However, although urinary AD7c-NTP was closely related to cognitive dysfunction in patients with FED, this relationship did not hold in RMDD patients.

Declaration of patient consent

The authors certify that they have explained the study protocol to all subjects who met the inclusion criteria, and clinical information began to be collected after all subjects agreed to participate in the study and signed the consent forms. The subjects understand that their names and initials will not be published and due efforts will be made to conceal their identity.

Financial support and sponsorship

This research was supported by the National Natural Science Foundation of China (Grant No. 81371494).

Conflicts of interest

There are no conflicts of interest.