Tissue-Specific Oxysterols as Predictors of Antidepressant (Escitalopram) Treatment Response in Patients With Major Depressive Disorder

Abstract

Background

There is growing evidence that disturbances in cholesterol metabolism may be involved in major depressive disorder (MDD). However, it is not known if cholesterol metabolites present in the brain and periphery can be used to diagnose and predict an MDD patient’s response to antidepressant treatment.

Methods

A total of 176 subjects (85 patients with MDD and 91 healthy control subjects) were included in this study. The expression of peripheral and brain-specific oxysterols and related gene polymorphisms were investigated in all subjects. The severity of depression was measured using the 17-item Hamilton Depression Rating Scale, 16-item Quick Inventory of Depressive Symptoms–Self-Report, and Patient Health Questionnaire-9 for all patients with MDD before and after 12 weeks of antidepressant treatment.

Results

Patients with MDD expressed higher plasma levels of 24(S)-hydroxycholesterol (24OHC) (mainly secreted from the brain) compared with healthy control subjects, and the higher levels of 24OHC were associated with 24OHC synthetase (CYP46A1) gene polymorphisms. In patients with MDD, an improved response to the 12-week antidepressant treatment was associated with a reduction of both 24OHC and 27OHC (mainly secreted from the peripheral system) levels relative to baseline levels. Nonresponders exhibited increased levels of oxysterols at the end of treatment compared with baseline. The superior reduction in oxysterol levels correlated with better outcomes from the antidepressant treatment.

Conclusions

These data suggest a potential role for oxysterols as diagnostic and treatment response–related indicators for MDD.

A disruption in cholesterol metabolism has been implicated in the development of major depressive disorder (MDD) (1, 2, 3). Synthesis and metabolism of cholesterol are relatively independent in the brain and peripheral system due to the blood-brain barrier (BBB) (4). However, it has not been established if and how the two tissue-specific cholesterols are involved in the pathogenesis of MDD and if these metabolites are good indicators to predict a patient’s response to antidepressant treatment.

Previous studies demonstrated that peripheral cholesterol levels might serve as a biological marker for MDD (5, 6, 7). However, the association between circulating cholesterol levels and MDD remains controversial. For example, some studies showed that lower circulating cholesterol is related to a higher risk of MDD (8, 9, 10), but other studies demonstrated opposite results (3,11). Furthermore, an increase in levels of total circulating cholesterols induced by antidepressant administration or electroconvulsive therapy was also reported in patients with MDD (12,13). Meanwhile, a different study reported on lower levels of cholesterol following antidepressant treatment (14). More importantly, hypocholesterolemic drugs reduced the risk of MDD (15). Taken together, the evidence suggests an important role for cholesterol in MDD. It is hypothesized that the contradictory reports of circulating cholesterol levels and their association with MDD might be related to tissue-specific cholesterol, metabolism, and genetic and environmental factors (16, 17, 18, 19). Therefore, to understand the impact of cholesterol on MDD, there is an urgent need to investigate the central or peripheral cholesterol metabolism.

As the major circulating oxysterol in the bloodstream, peripheral cholesterol can be oxidized by cytochrome P450 CYP27A1 into 27-hydroxycholesterol (27OHC), which can cross the BBB to gain entry into the brain by diffusion (20). Both CYP27A1 and 27OHC have been implicated in several neuropsychiatric diseases (20, 21, 22, 23). Changes in the circulating levels of 27OHC are induced by various insults such as oxidative stress and hypercholesteremia (24,25). An increase in 27OHC levels might promote oxidative stress–induced cell damage and contribute to the development of neuropsychiatric diseases (26,27).

The brain is the most cholesterol-rich organ and contains 25% of the body’s cholesterol content (28). Because cholesterol cannot cross the BBB, the brain and peripheral system are able to independently synthesize and metabolize cholesterol (4). For example, brain cholesterol is mainly oxidized into 24(S)-hydroxycholesterol (24OHC), which is catalyzed by the brain-specific enzyme CYP46A1 and released into the bloodstream (29). More than 90% of plasma 24OHC can be attributed to the brain (24,25); therefore, circulating 24OHC reflects brain-specific metabolized cholesterol (4). The source of residual 24OHC in the blood has not been determined but may result from the conversion of cholesterol by CYP27A1 to produce 24OHC, 25OHC, and 27OHC (30). It has been reported that the ratio of 24OHC to total cholesterol in plasma is similar to that in cerebrospinal fluid (r² = 0.94) (31). 24OHC plays a critical role in the physiological cholesterol homeostasis within the brain, and abnormal changes in 24OHC levels are associated with diseases such as Alzheimer's disease (32, 33, 34), amyotrophic lateral sclerosis (35), Parkinson's disease (14), and autism spectrum disorders (36). Our previous study also indicated an important role of oxysterols in schizophrenia (37). However, there have been limited studies on brain cholesterol metabolism in MDD (5,10). Importantly, a human postmortem study showed a significant increase in 24OHC in the brain of patients with MDD who had committed suicide (38), while another postmortem study indicated no significant change in 24OHC levels associated with the cortex of depressive patients (39). Together, both postmortem human studies proposed an important role for brain cholesterol turnover in MDD, which might differ from changes in peripheral cholesterol levels. The aim of this study was to determine if central and/or peripheral originated oxysterols are involved in the pathogenesis of MDD, and if the levels are able to predict the efficacy and outcome of 12-week antidepressant treatment in patients with MDD.

Methods and Materials

Participants

The participants in this study were part of a multicenter, randomized clinical trial (ChiCTR-OOC-17012566). A total of 176 participants were recruited from the Beijing Anding Hospital of Capital Medical University in China from May 2017 to December 2020 (40). The subjects comprised 85 patients with MDD and 91 age- and sex-matched healthy control (HC) subjects. The study was approved by the Independent Ethics Committee of Beijing Anding Hospital (no. 2017-24) (14). All participants completed written informed consent before being included in the study. Figure 1 represents an overview of the study undertaken.

Figure 1.

An overview of participant selection and treatment protocols. HAMD-17, 17-item Hamilton Depression Rating Scale; HC, healthy control; MDD, major depressive disorder; QIDS-SR16, 16-item Quick Inventory of Depressive Symptoms–Self-Report.

The enrolled patients met the following criteria (14): 1) was an outpatient between 18 and 65 years of age; 2) met the DSM-IV criteria for MDD; and 3) scored 11 or greater on the 16-item Quick Inventory of Depressive Symptoms–Self-Report (QIDS-SR16) (41), and scored 14 or greater on the 17-item Hamilton Depression Rating Scale (HAMD-17) (42). Clinical evaluations were performed through interviews conducted by at least 2 experienced research psychiatrists. The exclusion criteria were the following: 1) had a lifetime history of other psychiatric disorders such as bipolar, schizophrenia, or schizoaffective; 2) had used antidepressant medication over a period of 7 days or longer 14 days prior to their recruitment; 3) had a history of drug or alcohol abuse; 4) was pregnant or breastfeeding; 5) had a serious suicide risk (score of ≥3 on item 3 of the HAMD-17); 6) had severe chronic diseases with ongoing treatment; and 7) had significant medical conditions and unstable psychiatric condition.

The HC subjects were recruited through advertisements and were deemed as mentally healthy based on the criteria of any DSM-IV Axis I psychiatric disorder. In addition, HC participants were excluded if they: 1) had a history of any type of mental disorder; 2) had significant medical conditions; and 3) were pregnant or breastfeeding.

Clinical Treatment and Evaluation

After enrollment, all patients were treated with escitalopram, a commonly used selective serotonin reuptake inhibitor, for 12-weeks to evaluate the effectiveness of the antidepressant. The starting dose of escitalopram was 5 to 10 mg/day based on the individual clinical evaluation, and the average dose was 15.83 ± 4.12 mg/day by the end of the 12th week. During the 12-week treatment period, the severity of depression was assessed with the HAMD-17, QIDS-SR16, and Patient Health Questionnaire-9 (43) at the 4th, 8th, and 12th weeks after initiating treatment. An additional evaluation on the severity of depression was performed using the QIDS-SR16 at the second week. Two clinically experienced psychiatrists who were not informed of the antidepressant treatment assessed the patients’ symptoms. Before subject enrollment, all psychiatrists were trained on the consistency of the scale. The clinical assessment was performed and drug administration recorded at each visit. In addition, all subjects completed self-assessment and mood mapping on the mobile app every day. Patients with MDD who showed a good response to the treatment (a reduced HAMD-17 score by at least 50% posttreatment compared with the initial assessment) were recorded as responders (44,45) and others were identified as nonresponders.

Sample Collection

Plasma samples were collected before treatment (baseline) and at the 2nd and 12th weeks after initiating treatment. In brief, blood samples were collected in EDTA tubes. For DNA isolation, the blood samples were stored at −80 °C until required. For plasma preparation, blood was centrifuged at 1500g for 10 minutes and the plasma was liquated and stored at −80 °C.

Oxysterol Analysis

Plasma cholesterol levels were assessed in the Clinical Laboratory of Beijing Anding Hospital using enzymatic methods (Modular; Roche Diagnostics) as described by Hummel et al. (46). Plasma oxysterol levels were detected using high-performance liquid chromatography–mass spectrometry as described previously (37). In general, 50 μL of plasma or standard sample (with 5% bovine serum albumin) was mixed with 200 μL of acidic buffer solution (50 mM ammonium acetate, 1% formic acid, pH 3) and 1 mL methyl tert-butyl ether in a microcentrifuge tube, while 100 ng D5/D7 deuterium cholesterol was spiked in as an internal standard. The supernatant was harvested after fully mixed and dried down with nitrogen at 30 °C. For the derivatization, the dried samples were reacted with 12.6 g/L N,N′-diisopropylcarbodiimide, 12.4 g/L nicotinamide, and 12.2 g/L 4-dimethylaminopyridine in 50 μL of chloroform solution at 35 °C for 2 hours. The solution was then dried and dissolved in 100 μL of methanol for high-performance liquid chromatography–mass spectrometry analysis.

Genotyping

Genomic DNA was extracted from whole blood using standard methods (QIAGEN). The ratio of optical density 260/optical density 280 was calculated to evaluate the quality of the extracted DNA. DNA extracted from one particular patient was excluded due to the poor quality of the extracted genomic DNA. DNA samples that passed quality control were submitted for genotyping. Genotyping was carried out with the Capital Biotechnology Precision Medicine Research Array Kit (Thermo Fisher Scientific), which is a customized chip based on the Axiom 2.0 platform (Thermo Fisher Scientific). This microarray contains 787,400 single nucleotide polymorphisms (SNPs) and includes 50,000 novel markers specifically for East and South Asian populations, particularly Chinese, based on the human genome version 19 (Genome Reference Consortium Human Build 37). A series of different criteria were used to ensure quality control, including call rate (≥95%), minor allele frequency (≥0.01), and Hardy-Weinberg equilibrium p values (≥1 × 10⁻⁶).

Statistical Analysis

Using α = 0.05 and power = 0.90 (Power Analysis and Sample Size, PASS), we projected that the sample size needed to detect differences in oxysterol levels between patients with MDD and HC subjects would be 78 (n = 39 per group). All variables were analyzed using SAS version 9.4 (SAS Institute). The variables were expressed as mean ± SD. Differences in the variables between the HC and MDD groups were analyzed using χ² test, independent-sample t test, or Wilcoxon rank sum test. Comparison of the variables between HC subjects and MDD responders and nonresponders at baseline was performed using one-way analysis of variance followed by post hoc Bonferroni multiple comparison. Changes in oxysterol levels at baseline and the 2nd and 12th weeks were analyzed via paired t test. The corrected p value < .0167 (.05/3) was considered statistically significant after Bonferroni correction for multiple comparison in oxysterols and cholesterol. Multiple linear regression analysis was used to assess the relationship between changes in oxysterol levels and severity of depression in patients with MDD from baseline to the endpoint of treatment with age, sex, and body mass index as covariates. Two-way repeated-measures analysis of variance with Bonferroni adjustment was used to compare the oxysterol levels between responder and nonresponder groups. The differences in clinical assessments and oxysterol levels between genotypes were analyzed via independent sample t test. The receiver-operating characteristic (ROC) curve and the area under the ROC curve were used to evaluate the diagnosis value of oxysterol levels in patients with MDD. The genotyping analysis was assessed with PLINK 1.9 (https://www.cog-genomics.org/plink/1.9/). The Hardy-Weinberg equilibrium was conducted by χ² test for each SNP of every participant. The multivariate logistic regression analysis was used to evaluate the association between SNPs and MDD susceptibility or plasma 24OHC levels at baseline, adjusting for age and sex. The multiple-corrected p value of <1.33 × 10⁻⁷ was considered statistically significant after Bonferroni correction for the gene analysis. In other cases, the level of significance (p value) of <.05 was considered statistically significant.

Results

Demographic and Clinical Characteristics

The sociodemographic and clinical characteristics of the participants are presented in Table 1. There were no significant differences for all demographic data between patients with MDD and HC subjects, despite the MDD group showing a trend of higher body mass index than HC subjects.

Table 1.

Demographic and Clinical Characteristics

	HC, n = 91	MDD, n = 85	Statistic	p
Demographic Data
Age, years	29.31 (6.98)	28.95 (7.56)	t174 = 0.336	.737
Sex, female/male	57/34	53/32	χ21 = 0.002	.969
Body mass index	21.78 (3.87)	22.68 (3.45)	t174 = −1.625	.106
Clinical Characteristics
Family history, MDD, yes/no	0/91	20/65	χ21 = 24.157	<.001
Onset age, years	NA	25.08 (7.18)	NA	NA
Duration of illness, years	NA	2.91 (4.04)	NA	NA
Clinical Assessments
HAMD-17 score at baseline	NA	20.60 (4.26)	NA	NA
HAMD-17 score at endpoint	NA	7.63 (6.28)	NA	NA
QIDS-SR16 score at baseline	NA	15.21 (3.38)	NA	NA
QIDS-SR16 score at endpoint	NA	6.91 (4.63)	NA	NA
PHQ-9 score at baseline	NA	16.55 (4.34)	NA	NA
PHQ-9 score at endpoint	NA	6.41 (5.99)	NA	NA

Values are mean (SD) or n.

HAMD-17, 17-item Hamilton Depression Rating Scale; HC, healthy control; MDD, major depressive disorder; NA, not available; PHQ-9, Patient Health Questionnaire-9; QIDS-SR16, 16-item Quick Inventory of Depressive Symptoms–Self-Report.

Higher Levels of Cholesterol and 24OHC in Patients With MDD Compared With HC Subjects

As shown in Table 2, plasma cholesterol levels were significantly higher in patients with MDD compared with HC subjects (p < .001). At same time, plasma 24OHC levels were also higher in patients with MDD compared with HC subjects (p < .001). Hence, the 24OHC/cholesterol ratio was significantly higher in patients with MDD than control subjects (p = .009). On the other hand, no significant difference was observed in plasma 27OHC levels between patients with MDD and control subjects (p = .799). There was also no difference in the 27OHC/cholesterol ratios between patients and control subjects (p = .080).

Table 2.

Comparison of Cholesterol Metabolism Between Patients With MDD and HC Subjects and Patients Prior to and Post Antidepressant Treatment

	HC	MDD						HC vs. MDD (Baseline)
	Baseline	Baseline (W0)	W2	W12	p1	p2	p3	Statistic	p
24OHC, ng/mL	33.11 (11.49)	44.08 (17.31)	43.54 (17.41)	42.66 (16.93)	.644	.364	.712	t174 = −4.990	<.001a
27OHC, ng/mL	58.38 (38.53)	57.08 (28.20)	55.54 (26.70)	53.91 (26.77)	.496	.047	.396	z = 0.255b	.799
Cholesterol, mmol/L	4.27 (0.91)	4.85 (1.15)	4.70 (0.99)	4.82 (0.95)	.114	.747	.263	t172 = −3.759	<.001a
24OHC/Cholesterol Ratio, mg/g	20.71 (6.80)	23.84 (8.71)	24.63 (10.57)	23.26 (9.02)	.348	.572	.319	t172 = −2.649	.009a
27OHC/Cholesterol Ratio, mg/g	37.21 (25.81)	31.38 (16.46)	31.87 (16.10)	29.81 (15.33)	.398	.124	.118	z172 = 1.760b	.080

Values are mean (SD). A corrected p value < .0167 (.05/3) was considered statistically significant after Bonferroni correction. p1: W0 vs. W2; p2: W0 vs. W12; p3: W2 vs. W12.

24OHC, 24(S)-hydroxycholesterol; 27OHC, 27-hydroxycholesterol; HC, healthy control; MDD, major depressive disorder; W, week.

^aStatistically significant differences.

^bWilcoxon rank sum test.

To examine the disease specificity of plasma 24OHC levels in MDD, we subsequently performed an ROC curve analysis. The area under the ROC curve obtained by the ROC analysis was 0.783 (95% CI, 0.717–0.849) when the indicators of age, sex, and body mass index were combined, denoting the utility of plasma 24OHC levels to differentiate patients with MDD from HC subjects. The best area under the ROC curve result showed a sensitivity of 74.7% and specificity of 72.5%, suggesting that plasma 24OHC levels might be a potential diagnostic biomarker of MDD.

CYP46A1 Gene Polymorphisms and 24OHC Levels in Patients With MDD and HC Subjects

To explore the mechanism of increase in 24OHC levels in patients with MDD, we conducted a gene polymorphisms analysis in subjects. Unfortunately, we found no significant difference in SNPs (n = 375,184) between patients with MDD and HC subjects (all ps > 1.33 × 10⁻⁷ [.05/375,184] after Bonferroni correction) (Table S1). Considering that CYP46A1 gene polymorphisms might be an important factor to 24OHC levels, we analyzed the SNPs in CYP46A1 (n = 100) between patients with MDD and HC subjects to explore the linkage between 24OHC levels and oxysterol-related gene polymorphisms (Table S2). Our data showed 6 SNPs in the CYP46A1 gene potentially associated with susceptibility to MDD (Table 3 and Table S2); however, these differences did not reach statistical significance after Bonferroni correction (all ps > 5 × 10⁻⁴ [.05/100]). We conducted linear regression analyses on the 6 SNPs and levels of 24OHC to explore whether the disease-associated variants could account for the difference in 24OHC levels among individuals. Interestingly, we observed that an increase in the minor allele (CA) dose of rs35267715 was associated with increments in 24OHC levels (β = 5.045; 95% CI = 1.331–8.760; t = 2.662, p = .009) in HC subjects. Coincidentally, more patients with MDD carried the CA allele (56%) than HC subjects (46%) (Table 3). However, any association between the minor allele (CA) dose of rs35267715 and plasma 24OHC levels was not evident for patients with MDD (p > .05) (Figure S1).

Table 3.

Associations of CYP46A1 Gene Polymorphisms With the Susceptibility to MDD

SNP	Allelea	HCb	MDDb	MAFc	Additive Modeld		Dominant Modeld		Recessive Modeld
					OR (95% CI)	p	OR (95% CI)	p	OR (95% CI)	p
rs4905883	G/T	0/4/87	2/12/70	0.02/0.10	4.245 (1.39–12.96)	.011	4.500 (1.40–14.44)	.012	NA	NA
rs1957512	G/A	1/7/83	2/17/65	0.05/0.13	2.542 (1.15–5.64)	.022	3.062 (1.26–7.46)	.014	2.222 (0.19–25.36)	.520
rs1957513	T/C	1/7/83	2/16/66	0.05/0.12	2.387 (1.08–5.29)	.032	2.849 (1.16–6.98)	.022	2.222 (0.19–25.36)	.520
rs1957515	C/A, G, T	1/8/82	2/18/64	0.05/0.13	2.439 (1.13–5.26)	.023	2.873 (1.22–6.75)	.015	2.222 (0.19–25.36)	.520
rs6575744	G/C	1/7/83	2/16/66	0.05/0.12	2.387 (1.08–5.29)	.032	2.849 (1.16–6.98)	.022	2.222 (0.19–25.36)	.520
rs35267715	CA/C	6/36/49	14/33/37	0.26/0.36	1.571 (0.99–2.48)	.052	1.489 (0.81–2.72)	.196	2.976 (1.06–8.36)	.038

HC, healthy control; MAF, minor allele frequency; MDD, major depressive disorder; NA, not available; OR, odds ratio; SNP, single nucleotide polymorphism.

^aMinor/major allele.

^bRare homozygote/heterozygote/major homozygote.

^cHC subjects/patients with MDD.

^dLogistic regression with adjustment for age and sex.

Association of Changes in Cholesterol and Oxysterol Levels and Improvement in Depressive Symptoms During 12 Weeks of Antidepressant Treatment

A multiple linear regression analysis was used to assess the association between the changes in oxysterol levels and changes in depression severity after the 12-week antidepressant treatment. As shown in Table 4 and Figure 2, changes in oxysterol levels were significantly associated with reduced scores for the HAMD-17 (24OHC: p = .045; 27OHC: p = .018) and QIDS-SR16 (24OHC: p = .088; 27OHC: p = .048).

Table 4.

Associations of the Changes in Oxysterol Levels and Improvement of Depressive Symptoms After Antidepressant Treatment

	HAMD-17		QIDS-SR16		PHQ-9
	t	p	t	p	t	p
24OHC
Baseline	1.092	.278	1.14	.269	0.974	.333
Week 2 – baseline	0.220	.826	0.497	.621	0.176	.861
Week 12 – baseline	−2.048	.045a	−1.733	.088	−0.899	.372
27OHC
Baseline	1.495	.139	1.699	.094	0.629	.531
Week 2 – baseline	0.030	.976	0.72	.474	0.843	.402
Week 12 – baseline	−2.432	.018a	−2.011	.048a	−0.365	.716
Cholesterol
Baseline	1.081	.283	1.071	.288	1.157	.251
Week 2 – baseline	0.175	.861	0.671	.449	0.748	.457
Week 12 – baseline	−0.641	.524	−0.435	.665	−0.652	.517

24OHC, 24(S)-hydroxycholesterol; 27OHC, 27-hydroxycholesterol; HAMD-17, 17-item Hamilton Depression Rating Scale; PHQ-9, Patient Health Questionnaire-9; QIDS-SR16, 16-item Quick Inventory of Depressive Symptoms–Self-Report.

^aStatistically significant differences.

Figure 2.

Multiple regression analysis of oxysterols and improvement in depressive symptoms of patients with major depressive disorder. The regression curves of plasma (A) 24(S)-hydroxycholesterol (24OHC) and (B) 27-hydroxycholesterol (27OHC) levels and Hamilton Depression Rating Scale (HAMD) reduction rates in patients with major depressive disorder. n = 85.

Changes in Oxysterol Levels With Different Treatment Outcomes

To further clarify the relationship between oxysterol levels and the effectiveness of treatment, patients with MDD were subgrouped as responders (n = 62 [72.9%]) and nonresponders (n = 23 [27.1%]). However, a total of 15 patients were excluded from the analysis, as their 12-week blood samples were not available; hence, all further analysis is based on 51 responders and 19 nonresponders (Figure 3A, B). In general, 12 weeks of antidepressant treatment did not result in any significant changes in average total plasma cholesterol or oxysterol levels in patients with MDD (Table 2). However, at the end of the 12-week treatment, only the responders exhibited a significant reduction in both 24OHC and 27OHC levels when compared with baseline levels, as shown in Figure 3C and D. In contrast, the nonresponders demonstrated elevated 24OHC and 27OHC levels after completing the treatment (Figure 3E, F). Repeated-measures analysis of variance showed that there was no significant main effect for time (F₁ = 0.002, p = .961), group assignment (F₁ = 0.335, p = .565), or interaction (F₁ = 2.747, p = .102) in 24OHC levels. However, we found a significant interaction effect (F₁ = 8.498, p = .005) in 27OHC but no significant main effect for time (F₁ = 0.276, p = .601) or group assignment (F₁ = 0.072, p = .789).

Figure 3.

Different changes in depressive symptoms and oxysterol levels between the responders and nonresponders after antidepressant therapy. The different alteration of (A) Hamilton Depression Rating Scale (HAMD) and (B) Quick Inventory of Depressive Symptoms (QIDS) scores in the responders and nonresponders. The difference in levels of (C) 24(S)-hydroxycholesterol (24OHC) and (D) 27-hydroxycholesterol (27OHC) in responders, nonresponders, and healthy control (HC) subjects after antidepressant treatment. A contradictory change in both 24OHC and 27OHC levels after treatment was noted in responders and nonresponders. The reduction of (E) 24OHC and (F) 27OHC levels from baseline to week 12 in responders and nonresponders. ∗∗p < .01 and ∗∗∗∗p < .001 vs. HC subjects at baseline in panel (C); #p < .05 vs. responders at baseline in panel (C); ##p < .01 vs. responders at baseline in panel (D); and ∗p < .05 and ∗∗p < .01 vs. responders in panels (E) and (F), respectively. n = 51 in the responder group; n = 19 in the nonresponder group; n = 91 in the HC group.

Discussion

Given the previous association between cholesterol levels and the pathogenesis of MDD, this study was initiated to explore tissue-specific cholesterol metabolism in MDD and its relationship with the effectiveness of antidepressant treatment. We noted elevated brain cholesterol metabolism in patients with MDD, while changes in both brain and peripheral cholesterol metabolism were associated with antidepressant treatment. We demonstrated that only responders had reduced 24OHC and 27OHC levels compared with pretreatment or baseline levels. On the other hand, the opposite pattern was observed in the nonresponders after 12 weeks of antidepressant treatment.

First, we analyzed levels of the brain cholesterol metabolite 24OHC and peripheral oxysterol 27OHC in plasma to examine whether patients with MDD expressed any disruption in cholesterol metabolism in the brain and peripheral region. We found that plasma 24OHC levels, but not 27OHC levels, were significantly higher in patients with MDD compared with HC subjects. Because circulating levels of 24OHC might reflect general brain cholesterol hemostasis, the higher levels of plasma 24OHC could also be a result of neuronal damage–induced accumulation of cholesterol in the brain (29), or they could be involved in the pathogenesis of brain disorders. It is reasonable to hypothesize that elevated 24OHC plasma levels in patients with MDD might be related to neuropathology of MDD because brain atrophy and white matter lesions are often found in patients with MDD compared with HC subjects (47, 48, 49, 50, 51). Another possible regulator for brain-specific cholesterol levels is the activity of enzyme involved in cholesterol oxidization, such as CYP46A1 (52). In this study, HC subjects showed an increase in the minor allele (CA) dose of rs35267715 in the CYP46A1 gene, which was positively associated with increments in 24OHC levels. Interestingly, our data showed no such relationship between the CA allele with 24OHC levels in patients with MDD, even with a higher frequency of MDD CA carriers compared with HC subjects. While the impact of the CYP46A1 gene polymorphism rs35267715 on MDD pathology remains unresolved, the CYP46A1 gene rs754203 mutation has been reported to confer a high risk of Alzheimer’s disease as well as reduced circulating 24OHC levels in patients with Alzheimer’s disease (53,54). However, we were not able to identify any impact of rs754203 on 24OHC levels and risk of MDD (Table S2). Our data suggest that the higher levels of 24OHC in patients with MDD might be related to MDD brain pathologies, rather than to CYP46A1 polymorphisms. Another possible reason for the higher levels of 24OHC in MDD might be due to the higher levels of cholesterol in patients with MDD (Table 2). However, 27OHC levels or the ratio of 27OHC/cholesterol were similar in both patients with MDD and HC subjects; therefore, it seems unlikely that plasma cholesterol is responsible for the higher 24OHC level in patients with MDD.

While the underlying mechanism involved in the higher levels of circulating 24OHC in MDD remains unclear, studies have shown that enriched 24OHC levels might lead to many neurotoxic effects, such as inducing oxidative stress, inflammation, apoptosis, and excitatory toxicity (33,55, 56, 57, 58, 59). Furthermore, previous studies have shown that 24OHC might also serve as a potent positive allosteric modulator for the synaptic excitatory protein NMDA receptor to increase synaptic plasticity (60, 61, 62). In addition, elevated 24OHC levels might impair cholesterol synthesis and promote cholesterol excretion from the body by the liver X receptor (51). However, in our study, the higher levels of cholesterol observed in patients with MDD compared with control subjects suggests a limited role for liver X receptor.

Although strong evidence indicates that 27OHC is toxic to cells, including causing inflammatory reactions (53,59), promoting oxidative stress (27), and serving as an endogenous selective estrogen receptor modulator to link oxysterols and disease (63,64), we found no differences in 27OHC levels between MDD and HC subjects. It seems likely that cholesterol metabolism in the peripheral system might not contribute as much oxysterol as the central region toward the risk of experiencing MDD. A previous study indicated that 27OHC might promote less insult-induced neurotoxicity than 24OHC (65), suggesting a role for brain cholesterol metabolism in MDD. Further investigation into tissue-specific oxysterols together with a full panel of lipid metabolism in MDD might be important.

Another important and interesting finding from this study is that the changes to oxysterol levels were associated with an improvement in the depressive symptoms exhibited by patients with MDD. While all 3 clinical scores (HAMD-17, Patient Health Questionnaire-9, QIDS-SR16) were used in the evaluation of depression, only improvement of HAMD-17 scores showed significant association with changes in 24OHC levels. As previous studies indicated that the HAMD-17 is the most sensitive and reliable scale for depressive symptoms evaluation (66,67), a potential explanation of our result might be due to the individual feelings of the patients when performed, as the self-assessment scales varied greatly. In this study, we found that oxysterols, especially 27OHC, were reduced in the responders but increased in the nonresponders after antidepressant treatment when compared with levels at the baseline. We speculate that the relationship between oxysterols and the effectiveness of antidepressant treatment is dependent on the integrity of lipid rafts in the plasma membrane. It is known that localization of serotonin transporters into lipid raft microdomains is critical for serotonin uptake and is associated with the success of antidepressant treatment (34,68). Indeed, escitalopram is associated with the translocation of G protein alpha (Gsα) from lipid rafts (69,70). Both 24OHC and 27OHC are found in lipid raft microdomains. It is thought that an increase in 24OHC concentration could induce protein distribution within lipid raft microdomains to influence the lipid raft function (71), and an increase in 27OHC levels disrupted the lipid raft function through reducing the cholesterol density in the raft (45). These data imply that the increase in 27OHC after escitalopram treatment might have resulted from disrupting the integrity of lipid microdomains in the plasma membrane to dampen the antidepressant effect in the nonresponders. In addition, excess plasma 27OHC can cross the BBB into the brain, where it can activate nuclear factor-κB, oxidative stress, and inflammation, all of which are hallmarks of MDD (43,72, 73, 74, 75, 76).

This study has several limitations. First, we were able to complete the genotype analysis of only 84 patients, and future studies with larger sample sizes are needed to clarify the association between CYP46A1 polymorphisms and MDD as well as 24OHC levels in patients with MDD. Second, we only treated patients with MDD with escitalopram for 12 weeks, while a long treatment trial with follow-up investigation might provide more insight on tissue-specific oxysterols in patients with MDD (77). Third, we did not measure the levels of escitalopram and its metabolites in the blood to monitor the medication of patients. Last, the patients enrolled in this study mainly presented with mild forms of depression. Whether oxysterols play a role in patients with mild and severe symptoms should be investigated in the future.

To our knowledge, this is the first study to provide evidence of a relationship between MDD and plasma oxysterol levels following antidepressant treatment. Taken together, our findings suggest a potential role for oxysterols as diagnostic and treatment response–related markers for MDD and might represent targets for intervention in an attempt to slow disease progression and improve antidepressant treatment.