Serum plasminogen activator inhibitor-1 levels in patients with major depressive disorder vs. healthy controls: a systematic review and meta-analysis

Abstract Introduction Major depressive disorder (MDD) is a severe mental health condition that affects millions of people worldwide. Etiologically, several factors may play a role in its development. Previous studies have reported elevated plasminogen activator inhibitor-1 (PAI-1) levels in patients with depression, suggesting that PAI-1 levels might be linked to the etiology of MDD. Methods We systematically searched the following online databases: MEDLINE, Scopus, and Web of Science up to September 10, 2020, to identify studies in which PAI-1 levels were reported in subjects with MDD. Subsequently we used RevMan 5.3 to perform a meta-analysis of data extracted from the included studies using Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) and PICO criteria for the search and analysis. Results Six studies that reported mean ± standard deviation (SD) were included in the analysis, with a total of 507 MDD patients and 3,453 controls. The overall standardized mean difference (SMD) was 0.27 (95% confidence interval [95% CI] 0.01-0.53). PAI-1 serum levels were 0.27 SDs higher in MDD patients than in controls. The test for overall effect was significant (z = 2.04, p = 0.04). Substantial heterogeneity was detected among the studies, demonstrated by the inconsistency test (I2 = 72%) and the chi-square test (χ2 = 18.32; p = 0.003). Conclusion This systematic review and meta-analysis showed that MDD might be related to elevated PAI-1 levels. We propose larger prospective clinical studies to further investigate this clinical correlation and validate the clinical significance of these observations.


Introduction
Major depressive disorder (MDD) is a primary cause of morbidity and significantly affects the productivity of society. 1 The lifetime prevalence of MDD is approximately 10%, 2 imposing an economic burden of almost 92 billion Euros annually in Europe. 2 MDD impacts personal and family life negatively and is considered a risk factor for severe disorders, for example, Alzheimer's dementia, 3 cardiovascular illness, 4 and dependence on alcohol and other drugs.
MDD affects emotional functioning and quality of life, 5 and therapy-resistant MDD is associated with high suicide incidence. 6The etiology is heterogeneous and Major depressive disorder (MDD) is a severe mental health condition that affects millions of clinical presentation, course of illness, and treatment response can differ considerably.][9][10] The exact underlying etiological pathomechanisms remain unknown and treatments are not always effective.[13][14] Tissue plasminogen activator (tPA) is a major plasminogen activator that converts inactive plasminogen into active plasmin.Active plasmin breaks down fibrin clots, helping to restore vascular patency. 15AI-1 is a critical inhibitor of tPA, 14 and it was reported that low tPA levels are related to increased levels of its inhibitor PAI-1.14 The mechanism by which low tPA levels are related to MDD 16 might involve brain-derived neurotrophic factor (BDNF).The precurser of brainderived neurotrophic factor (proBDNF) is proteolytically cleaved into mature BDNF (mBDNF) by tPA and by plasmin (the end product of tPA) 16 and promotes neuronal apoptosis and long-term depression, while mBDNF has anti-apoptotic properties and favors longterm potentiation. 17Moreover, physical exercise has antidepressant effects and increases tPA and mBDNF levels.12 The inability to convert proBDNF into mBDNF is associated with depression pathogenesis and, consequently, mBDNF is implicated in the mechanism of action of antidepressants.18 Figure 1 shows the association between PAI-1 and MDD as different factors affecting PAI-1, including pleiotropic compounds, sleep disturbances, cortisol dysregulation, tPA level, hypothalamus-pituitarygonadal (HPG)-axis dysregulation, and inflammation.
The PAI-1 inhibits activation of plasminogen into plasmin which in turn leads to inhibition of conversion proBDNF into its mature form, mBDNF.Therefore, PAI-1 enhances apoptosis, spine retraction, and depression.
Figure 2 shows the association between PAI-1 and mBDNF.HPG-axis dysregulation, increased systemic inflammation, cortisol level, sleep disurbance, and obesity all increase PAI-1 levels.Elevated PAI-1 levels can cause MDD by inhibiting conversion of pro-BDNF into mBDNF.
Considering the reported data on the effects of tPAI on MDD pathophysiology and treatment resistance (Figure 1) and the lack of strong evidence on this topic, the authors decided to conduct a systematic review of existing studies in this field to provide a holistic, comprehensive conclusion on the impact of PAI-1 levels on depression.This study aims to examine and systematically review and analyze the literature and the reported results related to the impacts of PAI-1 levels on the pathophysiology of MDD and treatment options.

Data sources and searches
We systematically searched the MEDLINE (through PubMed), Scopus, and Web of Science online databases up to September 10, 2020, in order to identify studies that assess PAI-1 levels in MDD subjects using the following MeSH terms: ("SERPINE1" OR "Serpin E1" OR "PAI-1" OR "Type 1 Plasminogen Activator Inhibitor" OR "Plasminogen Activator Inhibitor-1") AND ("depression" OR "depressive").We included the studies in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA).We applied no study design or year of publication search filters.We

Data extraction
Data extraction was conducted by two independent authors.We extracted the summary data and statistical data to a Microsoft Excel spreadsheet template.More than one author checked the data extracted.The details of the searches and selection and inclusion of the studies are summarized in the following PRISMA diagram (Figure 3).

Study selection
Studies reporting PAI-1 levels in depressed human subjects published in English or German in international peer-reviewed journals were included.
Studies reporting mean ± standard deviation (SD) were included in the quantitative analysis.We excluded studies that reported other variables such as correlation or regression coefficients or that did not report a clear relationship between MDD and PAI-1, or that did not include a precise definition of depression.
We also excluded studies of psychiatric disorders not related to MDD, studies assessing the relation between genetics and depressive disorders, or animal studies reporting PAI-1.

Quality assessment of the included studies
The Studies (NIH-SQAT) was also used.This tool consists of 14 questions and allows three overall ratings according to the total score; "good", "fair", or "poor".

Data analysis
RevMan 5.3 was used in the meta-analysis of data. 19Five studies [20][21][22][23][24] reported mean ± SD, and one study 25 reported mean and range, so these data were transformed to mean and SD values using the method described by McGrath et al. 19 Eskandari et al. 20 reported two measurements of PAI-1 performed at 8:00 am and 8:00 pm; the average of both was calculated.Malan et al. 21reported the measurements of men and women separately; so the average of both measurements was taken, to be consistent with the other studies.SD was obtained from a 95% confidence interval (95%CI) using the method described in the Cochrane Handbook for Systematic Reviews of Interventions. 22

Sensitivity analysis
To examine the effect of a single study on the overall effect, a leave-one-out analysis was conducted.We also excluded an outlying study to examine the pooled effect and to account for heterogeneity.

Publication bias analysis
Egger's test and a funnel plot were employed to assess publication bias in the included studies.

Study characteristics
Following screening, a total of 34 studies were assessed for eligibility and six studies were chosen for inclusion in the meta-analysis.Detailed methodology and reasons for exclusion are presented in the PRISMA flow diagram (Figure 3).The characteristics of included studies and clinical data for both MDD and healthy control groups are presented in Tables 1, 2, and 3.The serum PAI-1 values are shown in Table 4.

Quality assessment
Following the NOS scoring categories, two studies were of high quality and four studies were of moderate quality (Table 5).According to the NIH-SQAT guidelines and after interrater consensus, two 21,23 of the included studies were considered to be of good quality (scored 7) and four 20,[24][25][26] were of moderate quality (scored 6).
Records identified from:

Sensitivity analysis
A leave-one-out analysis revealed that no single study affected the results significantly, as the SMD was still within the CI of the overall results following removal of each study.Eliminating some studies shifted the value to non-significant, where the CI intersects with 0, as shown in Figure 5.
We tried incorporating the 8:00 am measurements taken by Eskandari et al., 20 to be consistent with other studies instead of averaging the 8:00 am and 8:00 pm measurements, but doing so did not remarkably affect the result (SMD 0.28, 95%CI 0.02-0.55,p = 0.04).
Eliminating Pan et al. 23 reduces the heterogeneity level from 74 to 11% and increases the overall effect (SMD 0.36, 95%CI 0.17-0.55,p = 0.0002).However, eliminating studies on the basis of heterogeneity is not recommended, since it may introduce bias (Cochrane Handbook Chapter 10, Section 10.10.3). 22
However, the results of Egger's test should be interpreted with care when a meta-analysis includes a small number of studies.

Discussion
Previous research has provided evidence that patients with MDD often have elevated serum PAI-1 levels. 11-14PAI-1 is known for its inhibitory action on tPA. 15,16Inhibition of tPA predisposes to depressive symptoms due to involvement of BDNF. 15,16Mature BDNF is proteolytically cleaved from proBNDF by tPA and plasmin, which is itself the end product of tPA. 17hereas proBDNF stimulates neuronal apoptosis and depression, mBDNF has anti-apoptotic properties. 17ysical exercise that has antidepressant effects causes elevation of tPA and mBDNF levels. 12Failure of the conversion of proBDNF into mBDNF is associated with the pathogenesis of MDD.Mature BDNF is even implicated in the mechanism of action of selective serotonin reuptake inhibitor antidepressants, such as escitalopram. 18Therefore, this study aimed to systematically examine and analyze the results reported in the literature about the impact of PAI-1 levels on depression.To the best of our knowledge, our study is the first systematic review and meta-analysis to assess the association between PAI-1 and MDD.
In this meta-analysis, six studies 20,21,[23][24][25][26] were included, with a total of 3,960 participants (507 MDD patients and 3,453 controls).According to the results  presented in the six studies included in the quantitative synthesis, serum PAI-1 levels were 0.27 SDs higher in MDD patients than in controls.
The six studies included in this meta-analysis focused on rating scales rather than a clinical diagnosis for defining depression.Furthermore, the six studies included did not all focus on the same subtype of depression, so that the criteria used for diagnosis were not homogenous.A large proportion of the patients with depression included in the studies suffered from cardiovascular diseases, such as coronary heart disease (CHD) in the study by Lahlou-Laforet et al. 26 and coronary artery disease (CAD) in the study by Schroeder et al. 25 Furthermore, the included studies focused on specific age groups.Besides, differences in episodes, duration of illness, and illness severity among patients with MDD might be a potential confounding factor.Furthermore, PAI-1 levels are modulated by angiotensin, glucocorticoids, and aldosterone.However, the levels of these factors were not consistently screened in the six studies included.
Jiang et al. highlighted that in their rodent depression data model, stress increases PAI-1 expression in the medial prefrontal cortex, and the hippocampus.Also, chronic escitalopram treatment downregulated PAI-1 expression in these brain subregions and decreased the active PAI-1 concentration in the serum, but not in the CSF in rodents, 9 while it did not affect the PAI-1 concentrations in the serum of MDD patients. 25,26ny factors are known to affect PAI-1 expression, including obesity, 2 sleep dysregulation, 27 and hypothalamus-pituitary-adrenal (HPA) axis dysregulation 4 (Figures 1 and 2).Also, female hormonal dysregulation affects PAI-1, which leads to an increase in the incidence of female depression. 23,28Some pleiotropic compounds also affect PAI-1 activity, including metformin, 29 resveratrol, and other antioxidants such as curcumin, ginko biloba, shikonin, and theaflavin by many different mechanisms. 13,30armacological treatments available nowadays are useful, but unfortunately, around 30-40% of patients are resistant to such treatments. 31,32Besides, antidepressants could cause side effects. 29Party et al. introduced a new model for depression using PAI knockout mice, which showed resistance to antidepressant therapy. 28pression is known to be associated with alteration of PAI-1 protein, blood coagulation, and fibrinolysis. 33 Measurements of PAI-1 in serum and CSF would probably provide more reliable results.Intake of antidepressants might interfere with the results, as antidepressant treatment might influence PAI-1 levels. 35For example, in the study by Jiang et al., 24 escitalopram treatment significantly decreased PAI-1 levels in serum, but not in CSF.It is currently unclear whether PAI-1 can cross the blood-brain barrier, 34 despite its presence in certain areas of the brain and the CSF. 33espite the intensive search of the current literature and careful data extraction, this review still has limitations.Some studies focused only on one gender: Eskandari et al. examined only females, 20 while Lahlou-Laforet examined only males. 26

Conclusion
In conclusion, this systematic review and metaanalysis revealed that elevated PAI-1 is associated with MDD, however heterogeneity, publication bias, and the limited number of studies should be taken into consideration when interpreting the result.More extensive prospective clinical studies are required to thoroughly examine its clinical correlation and validate the clinical significance of these observations.
exported the search results to EndNote X9.1 (Clarivate Analytics, https://clarivate.com/) to remove duplicate studies.Next, the studies were exported to an Excel spreadsheet and screened by title and abstract by three independent reviewers (KA, RA, MS).Any disagreements about inclusion or exclusion of the screened studies were resolved by a senior study member.
Standardizedmean difference (SMD) was used as an overall effect measure.The studies were weighted by inverse variance, and a random-effects model was chosen owing to the heterogeneity of the data.Inconsistency (I²), chi-square (χ²), and tau-square tests were performed to test for heterogeneity.A p-value less than 0.05 was considered statistically significant; I² > 50% was considered indicative of substantial heterogeneity in the studies.22

Figure 3 -
Figure 3 -Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram illustrating selection of studies to be included in the systematic review and meta-analysis.PAI-1 = plasminogen activator inhibitor-1; SD = standard deviation.

Figure 4 -
Figure 4 -Forest plot of plasminogen activator inhibitor-1 (PAI-1) levels in major depressive disorder (MDD) patients compared to control groups; the right side represents higher serum PAI-1 in MDD patients, or lower levels in controls, the left side represents lower levels in MDD patients, or higher levels in controls.95%CI = 95% confidence interval; df = degrees of freedom; I² = inconsistency test; IV = inverse variance.

Figure 6 -
Figure 6 -Funnel plot test for publication bias using the effect measure and the standard error.
Eskandari et al. even   focused only on premenopausal females.20 Several confounding factors were present in the Eskandari et al. study; patients continued their antidepressant medication, which might be a confounding factor.Lahlou-Laforet et al.'s study included confounding factors such as smoking, hypertension, triglyceride concentration, and body mass index. 26It is worth mentioning that each study focused on certain coagulation factors including PAI-1; the coagulation factors that were studied varied between the six studies.Furthermore, Malan et al. focused only on members of the black African race 21 and Pan et al. focused solely on the Chinese population, 23 while the other studies probably involved various races.Also, Malan et al. measured fasting PAI-1 levels,21 while the other studies did not mention whether measurements were in a fasting state or not.Furthermore, data heterogeneity and publication bias were detected, and the limited number of studies still poses a main concern.

Table 1 -
Characteristics of the studies included in the meta-analysis

Table 2 -
Characteristics of the populations in the studies included

Table 4 -
Mean serum PAI-1 levels used in the meta-analysis MDD = major depressive disorder; SD = standard deviation.The mean represents the serum PAI-1 levels measured in patients and controls.* PAI-1 activity.

Table 3 -
Basic clinical data for the subjects included in the meta-analysis CES-D = Center for Epidemiologic Studies Depression Scale (20-item); CHD = coronary heart disease; f/m = female/male.Data are presented as mean ± standard deviation.* Median (interquartile range). 22

Table 5 -
Newcastle-Ottawa-Scale for quality assessment of included studies