Alpha 1-antitrypsin as a potential biomarker for diagnosing major depressive disorder

Despite decades of intensive research on major depressive disorder (MDD), the pathogenesis of MDD is still unclear and the objectively diagnostic method remains unavailable. Therefore, we conducted this study to assess whether alpha 1-antitrypsin (AAT) could be a potential biomarker for diagnosing MDD. Here, the levels of AAT, liver function-related indicators, renal function-related indicators, blood lipids-related indicators, high sensitivity C-reactive protein, homocysteine and transferrin were detected. The orthogonal partial least-squares discriminant analysis (OPLS-DA) was used to nd the differential variables, Random Forest was used to identify the simplied biomarker panel, and receiver-operating characteristic (ROC) curve analysis was used to evaluate the diagnostic performance of the identied panel. In total, 86 MDD patients and 99 healthy controls (HCs) were recruited. Finally, we found nine differential variables between MDD patients and HCs, and a potential biomarker panel consisting of AAT, albumin (ALB) and apolipoprotein A1 (APOA) was identied. This panel could effectively separate MDD patients from HCs in two independent samples sets. The level of AAT was signicantly negatively correlated with HDRS score and improved after antidepressant treatment. Meanwhile, MDD patients with suicide idea or behavior had signicantly lower AAT levels compared to MDD patients without suicide idea or behavior. Our results suggested that AAT held the promise to become a potential biomarker for diagnosing MDD, and also might be a potential novel therapeutic target for MDD.


Introduction
Major depressive disorder (MDD) is a common mental disease with high morbidity, disability and recurrence rates in clinical practice. Due to the increasing social pressure, the incidence rate of MDD is increasing year by year, and its object is gradually inclined to adolescents [1]. Restricted by various factors such as differences in regional development and cultural concepts, MDD also has the characteristics of low diagnosis rate and poor treatment effect. It has even become the main cause of suicide, bringing huge economic burden to individual, families and society [2][3][4]. At present, the diagnosis of MDD mainly relies on the subjective identi cation of symptom clusters, which unfortunately results in a considerable error rate [5]. Thus, it is urgently needed to identify disease biomarkers for objectively diagnosing MDD.
Many theories have been developed to explain the pathogenesis of MDD, such as neurotransmission de ciency [6], oxidative stress and immune system dysfunction [7,8] and gut microbiota alternation [9][10][11][12]. Luscher et al. proposed the central and causal role of GABAergic de cits in the etiology of MDD [6].
Previous studies reported that the disturbances in leukocyte function and/or leukocyte number could be potential biomarkers of MDD [7,8]. Our previous animal and human studies found that the disordered gut microbiota might have a causal role in the development of MDD [9,10]. These theories have made a great contribution to the prevention and treatment of MDD. But, none of these theories has been universally accepted. It means that some novel and comprehensive pathophysiologic mechanisms underlying the disorder are needed. Alpha 1-antitrypsin (AAT) is a single-chain glycoprotein mainly synthesized by liver cells and a positive acute phase protein (APP). It is the most abundant serine protease inhibitor in human blood. AAT could inhibit the synthesis and release of in ammatory mediators, and also inhibit the release of many proin ammatory factors, such as IL-1β, IL-6 and TNFα. Meanwhile, it could play an anti-apoptotic effect by eliminating cytotoxic substances, such as peroxides and free radicals. Our previous study found that AAT was decreased in serum of MDD patients [13]. Beiko et al. reported that anxiety and depression were common comorbidities in individuals with AAT de ciency [14]. Therefore, we conducted this study to further investigate the role of AAT in the onset and development of MDD, and explore whether or not AAT could be a potential biomarker for diagnosing MDD.

Patient enrollment
The MDD and healthy controls (HCs) subjects were recruited from the First A liated Hospital of Chongqing Medical University between October 2019 and January 2020. The Ethical Committee of Chongqing Medical University reviewed and approved this study. MDD subjects were independently diagnosed by two experienced psychiatrists according to DSM-IV criteria. Hamilton Rating Scale for Depression (HDRS) was used to evaluate the disease severity and Beck Scale for Suicide Ideation (BSI) was performed to assess the suicidal ideation. The MDD candidates with one or more confounding factors were excluded: preexisting physical or other mental disorders, younger than 18 or older than 65, female in pregnancy, lactation or menstruation, and/or illicit drug use. Meanwhile, HCs subjects were recruited from the Medical Examination Center of the same hospital. The HCs candidates with one or more confounding factors were excluded: any current or previous lifetime history of mental disorders, systemic medical disorder, female in pregnancy, lactation or menstruation, and/or illicit drug use.
Informed written consent was obtained from all the included subjects. The blood routine, liver functions and urine routine of the included subjects were without abnormal changing.

Samples collection and regrouping
Upon meeting all the above-mentioned inclusion and exclusion criteria, venipuncture was performed on site to collect the serum. The samples were transferred into the laboratory under low temperature and then stored at -80 °C until later analysis. Samples were collected at any time during regular business hours (9 am to 6 pm, Monday to Friday). We did not make any effect to restrict the timing for serum collection. Totally, serum samples from 99 HCs and 86 MDD patients were obtained. The detailed information of the included subjects was displayed in Table 1. These samples were randomly assigned into training set (69 HCs and 60 MDD patients) and testing set (30 HCs and 26 MDD patients). The 33 of the included MDD patients were received antidepressant therapy for one month, and then their serum samples were collected again.

Statistical analysis
Student's t-test, chi-square test, nonparametric Mann-Whitney U test, Pearson correlation analysis or paired t-test was used when appropriate. Levene's test was used to calculate the variances between the two groups; if the variances are not similar, the adjusted p-value was used here. The orthogonal partial least-squares discriminant analysis (OPLS-DA) was build using samples in training set, and the variables with variable importance in projection (VIP)>1.0 (equivalent to a p-value of less than 0.05) were viewed as the important metabolites responsible for discriminating MDD patients from HCs subjects. Meanwhile, the Student's t-test and Benjamini and Hochberg False Discovery Rate method were conducted to assess whether these identi ed important variables were still signi cantly different between the two groups. The variables with adjusted p-value <0.05 and VIP>1.0 were identi ed as the differential variables between HCs and MDD patients. Finally, the Random Forest was used to obtain a simpli ed biomarker panel from these identi ed differential variables. The receiver-operating characteristic (ROC) curve analysis was further used to evaluate the diagnostic performance of the identi ed simpli ed biomarker panel.

Discriminative model construction
The score plot of OPLS-DA model built with samples in training set showed that MDD patients and HCs subjects were clearly separated with little overlap (R2Y=0.71, Q2Y =0.60; Figure 1A), representing the strong explanatory power of the data. The positive values of R2Y and Q2Y indicated the robust differences between these two groups. Meanwhile, this model could effectively predict the samples in testing set: the T-predicted scatter plot showed that 26 of the 30 HCs subjects and 22 of the 26 MDD patients were correctly predicted ( Figure 1B), indicating the good predictive ability of the model. In addition, the 399-iteration permutation tests also suggested that the model was valid and stable.

Variables related with suicide idea or behavior
There were 56 MDD patients with suicide idea and 30 MDD patients without suicide idea. As shown in Figure 4A, compared to MDD patients without suicide idea, MDD patients with suicide idea had signi cantly higher HDRS scores (p=0.00015) and BSI scores (p=0.00003), and signi cantly lower AAT levels (p=0.00068); the ALB level and APOA level were similar between these two groups. There were 27 MDD patients with suicide behavior and 59 MDD patients without suicide behavior. As shown in Figure  4B, compared to MDD patients without suicide behavior, MDD patients with suicide behavior had signi cantly higher HDRS scores (p=0.02500) and BSI scores (p=2.51E-09), and signi cantly lower levels of AAT (p=0.0018), ALB (p=0.0097) and APOA (p=0.0160).

Differential variables changed after treatment
To study the changes of these identi ed differential variables after treatment, the levels of the variables before and after treatment in 33 MDD patients receiving antidepressant therapy were collected. The paired t-test was used to analyze the experimental data. As shown in Figure 5A, we found that eight of nine differential variables in MDD patients were improved after treatment. Moreover, compared to the levels before treatment, the levels of AAT, ALP, CHE, and APOA after treatment were signi cantly increased (p=0.0001, p=0.0081, p=0.0054, p=0.0201, respectively). Meanwhile, we found that both the HDRS scores (p=1.17E-15) and BSI scores (p=8.00E-05) were signi cantly decreased after treatment ( Figure 5B).

Discussion
Developing an objectively diagnostic method with high speci city and sensitivity for MDD has been proven to be a formidable and elusive task, although we and other researchers have completed many meaningful works [16][17][18][19]. In the present work, a potential biomarker panel consisting of AAT, ALB and APOA was identi ed. We examined the diagnostic performance of this panel in two independent sets: in training set, the application of this panel resulted in an AUC of 0.974 (sensitivity, 91.7%; speci city, 92.8%); in testing set, the AUC was 0.878 (sensitivity, 92.3%; speci city, 80.0%). Moreover, the levels of these potential biomarkers had been improved after treatment. These results demonstrated that this biomarker panel might be a "good" classi er of MDD patients and HCs, and our ndings could be helpful for future developing an objectively diagnostic method for MDD.
AAT has a variety of anti-in ammatory and tissue protection properties [20]. It could play an antiin ammatory effect by regulating various immune cells, such as neutrophil and lymphocytes [21]. Thus, AAT has an important role in many chronic diseases, such as liver cirrhosis and gastric cancer [20]. In recent years, the role of in ammation and oxidative stress in the pathogenesis of neuropsychiatric diseases has been recognized by many scholars [22,23]. Previous study showed that many MDD patients were accompanied by activation of the in ammatory response system (IRS), and then showed the compensatory immune regulatory response system (CIRS) activation [24]. As a positive acute phase response protein, AAT is an important component of CIRS. Therefore, there may be an important connection between AAT and MDD. Some studies reported that the level of AAT was decreased in serum of MDD patients compared to nondepressed subjects [25,26]. Another study found that the level of AAT was not signi cantly changed in plasma of depressed patients than in controls [27]. Using serum proteomics, we identi ed 74 differential proteins enriching in the acute phase response system, and AAT was found to be decreased in MDD patients than in HCs [13]. Meanwhile, Beiko et al. reported that anxiety and depression were common comorbidities in individuals with AAT de ciency [14]. In this study, we found that the AAT level in serum of MDD patients was the signi cantly decreased compared to HCs. This disparity may have resulted from differences in the demographic and clinical characteristics of the included MDD patients or HCs, sample sizes, and/or different races. However, the change of AAT level in MDD patients may be related to the phenomenon of oxidative stress during the onset of depression. MDD patients may consume large amounts of AAT for anti-in ammatory and immune regulation, then resulting in the decease of AAT level in serum.
There is a strong association between MDD and suicidal idea or behavior [28]. Here, we found that compared to MDD patients without suicidal idea or behavior, MDD patients with suicidal idea or behavior had the signi cantly lower average HDRS score. Our previous study found that suicidal behavior in MDD was associated with a more pronounced in ammatory phenotype [29]. In this study, we found that both suicidal idea and suicidal behavior were signi cantly negatively correlated with AAT, and suicidal behavior was also signi cantly negatively correlated with ALB and APOA. Moreover, the signi cantly lower average ALB and APOA levels were only found in MDD patients with suicidal behavior vs. MDD patients without suicidal behavior, not in MDD patients with suicidal idea vs. MDD patients without suicidal idea. These results might indicate that ALB and APOA were the two risk factors of MDD patients attempting suicide.
Several limitations of the present study should be taken into account when interpreting our ndings. Firstly, the sample size was relatively small, future studies with large sample size are needed to validate and support our conclusions. Secondly, the present study was conducted with adult HCs and MDD patients; whether or not the results are appropriate for children, adolescents or an exclusive sample of elderly patients is unknown. Thirdly, the subjects were recruited in a city of China (Chongqing); whether similar results would have been obtained in MDD patients from other places, such as Africa, Europe and North America, cannot be determined from this study.
In conclusion, our results showed that a potential biomarker panel consisting of AAT, ALB and APOA could yield high sensitivity and speci city in differentiating MDD patients from HCs. This was con rmed in two independent samples of MDD patients. The AAT was found to be signi cantly negatively correlated with HDRS score and was signi cantly improved after treatment, suggesting that it held the promise to become a potential biomarker for diagnosing MDD and also might a potential novel therapeutic target for MDD. Further research is needed to con rm the diagnostic performance of this panel in different populations.

Declarations
Author's contribution Jian-jun Chen developed the study concept. All authors contributed to the study design. Testing and data collection were performed by Shun-jie Bai, Kunxia Chen and Huili Bai. Shun-jie Bai, Jing Xie and Jian-jun Chen performed the data analysis and interpretation under the supervision of Huili Bai. Shun-jie Bai drafted the paper, and Jian-jun Chen and Jing Xie provided critical revisions. All authors approved the nal version of the paper for submission.