Major Depressive Disorder (MDD) is a highly prevalent and vastly complex clinical condition, which requires a multidimensional approach in its study (1–4). Several studies have highlighted that the risk of suffering from depression is related to cognitive patterns acquired during childhood, shaping the individual's ability to cope with daily life events during adulthood (5–9). In this regard, early environmental conditions, including maltreatment and a family history of depression, have been linked to the disorder (10, 11). There is also evidence of the mediating role of cognitive factors in the relationship between childhood maltreatment and subsequent adult psychopathology, including higher risk of developing depression (12–14).

Negative affective biases related to early life adversity (ELA) can increase the risk of developing MDD, which is known as the cognitive diathesis-stress model of depression (15). This cognitive model suggests that the negative attributional processes that the individual activates against life events are associated with the development of depressive symptomatology (16–18). Besides, the theory of early maladaptive schema (EMS) domains emphasizes the role of the cognitive dimension in the developing psychopathology, asserting that some patterns and tendencies acquired during childhood can represent an additional risk of suffering some kind of mental disorder (19, 20). EMS has been defined as pervasive and dysfunctional beliefs about the self and the relationship with others, which are classified into five domains: “Disconnection and Rejection” (DR), “Impaired Autonomy and Performance” (IAP), “Impaired Limits” (IL), “Other-directedness” (OD), and “Overvigilance and Inhibition” (OIN) (19, 21). There is a well-established link between developing EMS and the impossibility of satisfying psychological needs, such as manifesting autonomy or secure attachment, which is also related to parental care (12). For this point, there is evidence suggesting that individuals whose families have a history of depression are more likely to develop the disorder, given that they are exposed to adverse conditions created in the familiar context (22).

Biological mechanisms—epigenetics, for example—have been identified as crucial mediators of psychosocial factors and subsequent emerging risk for mental illness (23). The analysis of epigenetic marks associated with depression seems to be a promising path for the study of MDD. However, not all studies have shown differences in the methylation levels between depressive patients and controls (24–26), and there have been conflicting results, with either increased or decreased methylation within the promoter regions and its role in the pathogenesis of affective disorders (27–29). The methylation level within the brain-derived neurotrophic factor (BDNF, Gene ID: 627) has been one of the most studied epigenetic marks since this gene codes for a neurotrophin involved in the development and functioning of the nervous system, especially in neuronal growth, proliferation, and survival (25, 30, 31). The BDNF exon IV promoter region has been of particular interest as it is critical for activity-dependent transcription and includes several CpG sites liable to methylate (26, 27, 32). Despite these contrasting findings when reporting BDNF methylation in MDD, what seems clear is that ELA, including childhood maltreatment and neglect, is a key predictor for major depressive disorder, as it has been found to epigenetically affect critical behavioral systems (33).

The multidimensional nature of the disorder, and its wide phenotype spectrum has supposed a challenge for diagnosis. It has been reported that general practitioners correctly identify depression in about half of the cases. The rate of accurate diagnosis through clinical measurements has a wide variation between countries, with over-detection (false positives) as a latent problem (34), a situation that increase the need to further elucidate the nature of the disorder. Thus, this study intended to determine which psychosocial and cognitive variables (including early life adversity and maladaptive schema domains) and epigenetic marks (BDNF methylation levels at five CpG sites in promoter IV) could be useful to establish a multivariate model to differentiate individuals diagnosed with MDD from healthy controls.

A multifactorial logistic model was evaluated in the present study, including psychosocial, cognitive, and epigenetic factors associated with MDD. Regarding the family history of depression, recent evidence reports that familial risk increases the probability of individual lifetime depression (22, 46). Although it has been argued that the transgenerational effect of family risk may have a genetic background (47), the disorder's low heritability has encouraged the exploration of other hypotheses. It seems that familial risk has a strong link with early adverse conditions since high-risk families are more likely to experience a more significant number of challenging situations, which predicts the development of more severe depression (48). Concerning early adversity, there is a large body of evidence that experiencing abuse or maltreatment during childhood accounts as a risk factor in the development of psychopathology during adulthood (11, 49). The multivariate logistic regression in this study did not reveal a significant association between childhood maltreatment and MDD. This is explained by the mediating effect that we found of the cognitive schemas in the relationship of childhood adversity and the likelihood of depression. This finding is consistent across different studies confirming that cognitive factors could act as mediators between early maltreatment and the subsequent risk of developing psychopathologies (13, 14). This is comprehensible since, according to schema therapy (19), neglected children are at risk of developing EMS. Besides, some authors have pointed out that childhood adversity can indirectly predict depression through cognitive vulnerabilities, including dysfunctional schemas (50).

Additional evidence supporting the previous hypothesis states that the schema domains of impaired autonomy/performance and disconnection/rejection mediate the relationship between childhood maltreatment and depression (51). It is also consistent with prior research establishing that maladaptive schemas of impaired autonomy/performance, impaired limits and other-directedness have a strong association with MDD (5, 52–55). Schemas represent a cognitive dimension, which seems highly stable over time in depressed patients (20). Impaired autonomy and performance are related to the inability to cope in everyday life, which affects the individual in many dimensions, involving dependence and an underdeveloped self (21). Not surprisingly, it has been reported that the reduction of depressive symptoms in patients under treatment was strongly associated with a reduction in the punctuation of this schema (56). Impaired limits involve difficulties in setting internal limits, assuming responsibility or even setting long-term goals (57) and it have been identified as a predictor of depression severity (58). The other-directedness EMS implies subjugation, self-sacrifice and constant seeking for recognition, as well as a tendency to respect other's desires at the expense of one's own needs (57). This domain has also been highlighted for its relation with depressive symptoms (59) as well as for its mediating role between co-rumination and depression (60).

One of the biological processes underlying the etiology of major depression is epigenetics, which comprises molecular mechanisms, like DNA methylation, that modulate gene expression in response to extrinsic or intrinsic signals (61). In this study, we evaluated the methylation level at specific CpG sites in BDNF promoter IV. The brain-derived neurotrophic factor is a neurotrophin that has been widely associated with major depressive disorder since abnormalities in its expression produce dysfunction in circuits that compromise emotional and cognitive functions (24, 62, 63). Our association results for the CpG3 site in BDNF promoter IV suggest that individuals with elevated methylation levels are less likely to show a depressive phenotype. This finding contrasts with prior research establishing that individuals with depression have lower levels of BDNF than healthy controls (64, 65) which, hypothetically, would be related to higher levels of methylation in patients. However, results concerning the methylation of BDNF and its role in depression have been ambiguous. There is evidence showing higher methylation in some CpG regions of BDNF promoters I and IV of depressed patients than healthy controls (64), highlighting childhood adversity as a mediating factor of these differences (65). In contrast, other authors report that those regions can appear as either hypomethylated or hypermethylated in clinical populations (66). A study involving mothers and their newborns revealed that prenatal depressive symptoms predicted decreased BDNF IV DNA methylation in infants (28). Specifically, they found hypomethylation at the same CpG site assessed in the present study in newborns who were prenatally exposed to maternal stress compared to controls.

The differences found within that specific CpG site could imply changes in BDNF mRNA and protein expression since it is adjacent to the binding site for the transcription factor CREB (cAMP response element-binding), which modulates gene transcription via a DNA methylation-dependent mechanism (67, 68). Another study demonstrates that healthy individuals with a family history of depression exhibited higher peripheral BDNF levels (69), presumably suggesting hypomethylation in those individuals with higher familial risk. There is also evidence of DNA hypomethylation in fragments of BDNF IV in patients who have schizophrenia compared to controls (70).

It is important to note that CpG3 includes binding sites for the glucocorticoid receptor (GR), as shown on the PROMO platform (71). This gene is one of the most important for regulating the stress response through the Hypothalamic-Pituitary-Adrenal (HPA) axis and modulation of peripheral cortisol levels. Concerning depression, it has been shown that there is an alteration in the sensitivity of the GR, which causes control failure in the cortisol levels after activation of the HPA axis (72). Likewise, certain antidepressants directly affect glucocorticoid receptors, increasing their functionality and expression (73).

Our model underscores the importance of the individual's cognitive dimension and the echoes that it can produce at specific epigenetic marks to understand the depressive disorder better. Accordingly, to consider a complex disorder like major depression, we need a powerful kaleidoscope of factors that can help us to understand fractions and to gain a general perspective of a condition that involves many biological systems and, therefore, affects the individual in multiple dimensions in their daily life. Altogether, the results of our study suggest that a model including psychosocial, cognitive, and epigenetic factors could help differentiate depressive patients from healthy controls and, therefore, could contribute to clinical diagnosis. Indeed, it has been highlighted that biological and cognitive measurements will be crucial, beyond traditional symptom-based diagnosis, to subtyping and redefining psychiatric disorders (3, 74). Additionally, the inclusion of a biological variable, like DNA methylation of a critical gene previously related to depression, is an essential step toward future strategies for treatment and prevention. In the future, it would be benefit to explore association with other approaches, such as the endophenotypes described for depression and anxiety (75), including variables related to Behavioral Activation System (BAS) and Behavioral Inhibition System (BIS), which can act as moderators between depressive symptoms and live events (76). Also, adding functional analysis to clarify the effect of the methylation level at BDNF promoter IV, as well as testing the proposed model in samples from different populations could be very useful in future studies.

The data that support the findings of this study are available from the corresponding author upon request.

The studies involving human participants were reviewed and approved by Ethics Committees of the University of the Andes and the Montserrat clinic. The patients/participants provided their written informed consent to participate in this study.

MV, YG-M, EF, and ML: conceived and designed the research. MV, YG-M, EF, and WC: collected the data. MV, YG-M, SG-N, and WC: performed the analysis. MV, YG-M, EF, SG-N, and ML: wrote the paper. All authors contributed to the article and approved the submitted version.

This study was financed by the faculty of sciences, the vice-rectorate research of the University of los Andes and by a research grant (712.2015/908/120471250970) from Minciencias. They provided resources, but did not actively participate in any stage of the research.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.