Who will get depressed under major stress?


Depression doesn’t come from one gene, one life event, or one personality trait. That’s what makes it so hard to predict, prevent or treat effectively.

But new research suggests the power of a tool that uses a range of genetic information to predict a person’s chance of developing depression when they’re under intense stress.

The findings might help lead to a better understanding of the pathways that lead to depression.

While the tool is far from ready for common use on individuals, it does suggest the potential to personalize depression prevention, and identification of those who might be most vulnerable to stress or most resilient.

The study was done in a population of more than 5,200 people in the most stressful year of training for a medical career, called the intern year of residency. It’s described in a new paper in Nature Human Behaviour by a team from the University of Michigan.

A score based on many genetic factors

The team used a genetic risk-assessment tool called polygenic risk score. They constructed a polygenic risk score for major depressive disorder, or MDD-PRS, from widely available consortium and biobank data on the known associations between a person’s risk of depression, and variations throughout a person’s genome.

While genetics and stress are known to play a role in depression risk and onset, the new research helps reveal the way these factors interact.

The interns who had higher-than-average MDD-PRS scores were slightly more likely to be among the 3% of interns who showed signs of depression before their intern year started.

But by the end of the year, these high PRS subjects were much more likely to be among the 33% of interns who had developed depression.

On the other hand, the group with the lowest MDD-PRS scores were far less likely to show signs of depression throughout their intern year, suggesting that the scoring system could be used to identify those most likely to be resilient despite intense stress.

Interns as a model of depression and stress

The research team tested the predictive power of MDD-PRS on young doctors taking part in the Intern Health Study, which is led by Srijan Sen, M.D., Ph.D., the senior author of the new paper.

The Intern Health Study enrolls thousands of new physicians across the United States each year who agree to let the research team sample their DNA, and to answer surveys before they begin their intern year of medical training and several times during that intense year of long hours and high demands.

Sen and the study’s first author, U-M research specialist Yu Fang, combined data across millions of sites within the human genomes to construct the MDD-PRS, and looked to see how well a person’s “score” on this tool predicted with their scores on standard surveys of depressive symptoms.

They also assessed whether the MDD-PRS worked through known mechanisms to depression, such personal and family history, childhood experience or general temperament.

The result: the MDD-PRS they developed accurately predicted the chance that certain interns would develop depression symptoms while under stress.

“Interestingly, we found evidence that the association between MDD-PRS and depression is stronger in the presence of stress and that the additional predictive power of MDD-PRS under stress is largely independent of known risk factors for depression.” says Sen, who holds the Eisenberg Professorship in Depression and Neurosciences at U-M and is part of the U-M Department of Psychiatry, the Molecular and Behavioral Neuroscience Institute and the Depression Center.

“These findings further our understanding of how genomics and stress interact and suggest that further investigation of the genomics of stress response can uncover novel mechanisms that lead to depression.”

Notes of caution

The major limitation of the new study has to do with the source of the genetic information it’s based on. Because most research on the genetics of depression has been done in people of European ancestry, the MDD-PRS tool used in the study is specific to people of that background.

In fact, the tool failed to predict depression symptoms among groups of interns of south Asian or east Asian background who were also taking part in the Intern Health Study.

Also, the group of interns in the study was young, with an average age of 27, and had already graduated medical school and been accepted into a residency training program, making them not representative of the general population.

Despite these limitations, this test of the MDD-PRS suggests its potential use.

“We are optimistic that these findings will be transferred to other ethnic groups with improving multi-ethnic analysis techniques and more data collected from these populations,” says Fang. She also notes the predictive power of the MDD-PRS score to predict resilience is stronger than its ability to predict susceptibility. “We hope this will ease the worry that this tool might be misused to discriminate against at-risk individuals.”

More about the study

Sen, Fang and their colleagues used genetic linkages for depression identified through three major pools of DNA data: the Psychiatric Genomics Consortium, the UK Biobank and the commercial genetics company 23andMe.

Sen’s previous work has shown the rise in depression symptoms over the course of the intern year, and the importance of the personality trait known as neuroticism as well as of an intern’s own history of stressful childhood events or traumas, and any past diagnosis of depression.

Under low-stress conditions before intern year started, most of the genomic risk captured by MDD-PRS worked through these three established risk factors.

While genetics and stress are known to play a role in depression risk and onset, the new research helps reveal the way these factors interact.

However, while the overall predictive power of MDD-PRS increased significantly under the stress of intern year, the predictive power of the three established factors remained the same.

“This suggests that as-yet-undiscovered factors accounted for the greater link between MDD-PRS and depression under the high stress conditions, suggesting a possible pathway to learning more about those undiscovered factors,” says Sen.

“The ability of better predict who is most at risk for depression could allow us to better target our powerful interventions to prevent depression,” he concludes. “Studying individuals with a genomic protection against depression under stress can help us understand resilience”

In addition to Sen and Fang, the study’s authors include Laura Scott, Ph.D. and Peter Song, Ph.D. of the U-M School of Public Health, and Margit Burmeister, Ph.D., of the U-M Department of Human Genetics and MBNI.

Funding: The study was funded by the National Institute of Mental Health (MH101459)

Depression is a leading cause of disability worldwide1 and is expected to be the highest global burden of disease by 2030.2 

Despite efforts to improve interventions, prevalence is still increasing, especially in adolescence.3 Evidence suggests that depression during adolescence is associated with many concurrent and later psychological and social impairments.4,5

However, what is driving this increase in adolescent depression is still not clear. A greater understanding of the nature of adolescent depression and how to minimize it is crucial if we are to reduce this global burden.

There is evidence that depression should be viewed on a continuum6,7 because individuals with subthreshold depression8,9 and elevated levels of depression symptoms10 are also at risk of concurrent and later psychopathology.

Importantly, a similar pattern is also observed for those displaying consistently higher levels of depression symptoms over time.1115 Detailed longitudinal analysis provides an opportunity to further understand not only how depression may manifest differentially over time but also its etiology.

Toward this goal, researchers have identified trajectories of depression symptoms during adolescence and potential risk factors associated with those trajectories.16,17 Previous research suggests that adolescence is characterized by a clear increase in symptoms of depressed mood; however, trajectories vary within the population and differ by age at onset as well as duration and severity of symptoms.12,1618 

Evidence has shown that several less favorable trajectories of depression symptoms (ie, those with depression symptoms that start high and continue [childhood persistent], those that start low but increase over time [early-adult onset], or those that start high in early childhood but decline during adolescence and young adulthood [childhood limited]) are present across multiple populations and often associated with poorer outcomes compared with adolescents with low symptoms over time (ie, the stable-low trajectory).11,12,19,20 It may be possible to start targeting specific interventions and treatments for certain individuals by disentangling the risk factors (or combinations of risk factors) of different trajectories.

Identifying how different risk factors are associated with varying patterns of depressive mood could be important for understanding the etiology of depression and improving treatment. Risk factors such as sex,14,19 childhood psychopathology,11,20 parental mood,19,21 and early-life socioeconomic position11,22 are important factors of less favorable trajectories of depression symptoms. Stronger associations are typically observed for chronically high or increasing trajectories,11,1821 yet the evidence is not clear cut.

A study by Rice et al23 also found that polygenic risk for depression was associated with a late-adolescence–onset trajectory, implying that genetic liability may be a key component for less favorable trajectories at specific periods of development.24 

The evidence is less clear on what might affect trajectories limited to adolescence or childhood, although research suggests they could be reactive to more immediate stressors and events.11 For example, bullying is arguably among the strongest factors in adolescent and adult depression2527 and is most frequent and salient during certain periods of childhood and adolescent development. Therefore, bullying in childhood is likely to have immediate consequences.

However, the direction of this association is unknown, and there is evidence that bullying is also associated with depression throughout the life course, suggesting it could also be reflective of preexisting susceptibility.28 

Therefore, it is unclear whether the association of bullying with depression is time specific, dependent on other prior factors, or both.29

 Investigating how bullying affects differential trajectories of depressive symptoms across adolescent development could give insight into how and when the effect of this risk factor occurs. Bullying in childhood has also yet to be linked to different trajectories of depressive symptoms.

Trajectories of depression symptoms are likely to have a highly complex etiology composed of genetic and environmental influences. This is because behavioral phenotypes, such as depression, are unlikely to have purely direct genetic or environmental pathways.30 

Instead, it is more likely that an interplay of genetic-environmental correlations exists and the contribution of genetic or environmental factors may be greater for complex traits (such as depression).7,27,3133

 For example, stressful life events may cause more severe depression symptoms, but it is possible that genetically liable individuals may be more prone to stressful life events, thus making it hard to determine the direction of effects.

Therefore, while we cannot yet separate whether a risk factor operates through genetic or environmental mechanisms, examining both genetic and environmental risk factors could build better prediction models and provide a new understanding that could be translated into improved prevention and interventions.

To our knowledge, no studies have examined the longitudinal nature of trajectories of depression symptoms in adolescence or their associations with genetic and environmental risk factors in early childhood and adolescence.

Understanding whether different risk profiles are associated with specific manifestations of trajectories of depression symptoms may offer more precise opportunities to target interventions during certain periods. We hypothesized that the most high-risk trajectories (ie, childhood persistent) would be associated with a combination of genetic and environmental risk factors (possibly reflecting a complex genetic-environmental interplay and/or a greater genetic contribution, in which genetic liability is reinforced by environmental events).

However, given the ambiguity surrounding the antecedents of adolescent-limited and childhood-limited trajectories, there may be some specificity in these trajectories that could be more reflective of emotional reactions to recent negative environmental events, such as bullying.

MethodsStudy Sample

We used data from the Avon Longitudinal Study of Parents and Children (ALSPAC), a longitudinal cohort study that recruited pregnant women residing in Avon, United Kingdom, with expected delivery dates of April 1, 1991, to December 31, 1992.34,35 

The initial cohort consisted of 14 062 children. Ethical approval was obtained from the ALSPAC Ethics and Law Committee and the local research ethics committees. Participants provided written informed consent to the collection and use of these data. Data in this study were fully anonymized.

The study website contains details of the data that are available through a fully searchable data dictionary and variable search tool.36 This report follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.37Depression Symptoms

Self-reported depression symptoms were measured on 9 occasions using the Short Mood and Feelings Questionnaire (SMFQ)38 when participants were aged 10 to 24 years. The SMFQ is a 13-item questionnaire that measures the presence of depression symptoms in the last 2 weeks.

It was administered via mail or in clinics. Each item is scored from 0 to 2, resulting in a summed score from 0 to 26. The SMFQ correlates highly with clinical depression39,40 and has been used to explore trajectories of depression symptoms in other studies.41,42 Descriptive information can be found in eTable 1 in the Supplement.Risk Factors

Sex (male vs female) was identified from birth notifications around the time of delivery. The polygenic risk score (PRS) for depression symptoms was created using summary statistics from a recent genome-wide association study (GWAS) on depression symptoms.43 

The PRS was created by weighting the effect sizes of 120 422 single-nucleotide polymorphisms associated with depression symptoms from the initial GWAS at a P < .50 threshold.

The PRS was standardized to have a mean of 0 and a standard deviation of 1; thus, a higher PRS represents higher liability to depression symptoms. Maternal postnatal depression (yes vs no) was assessed at 8 weeks post partum, when the mother completed the Edinburgh Postnatal Depression Scale.44 

A cutoff score of 13 or higher, which indicates probable depression, was used.45,46 Partner cruelty to the mother (yes vs no) was assessed by the Family Adversity Index, which asked the mother whether her partner was abusive toward her when the child was aged 2 to 4 years.22 Childhood anxiety was measured at approximately age 8 years by asking the child’s main caregiver about the child’s general anxieties, using questions taken from the Development and Well-being Assessment.47 

A weighted summary score of 0 to 12 was created from 6 questions on childhood anxiety, with 12 indicating maximum anxiety. Childhood bullying was measured using the modified Bullying and Friendship Interview Schedule when the child visited a research clinic at approximately age 10 years.48 We used a binary variable (yes vs no) to assess bullying of the child in the last 6 months. Further information on these variables can be found in the eAppendix in the Supplement.Confounders

The following confounders were included based on previous literature that associated them with the risk factors and the depression trajectories41,42,45: maternal age at birth (in years), maternal socioeconomic status at birth (manual vs nonmanual occupation), maternal educational attainment at birth, parity (first born vs second born vs third born or more), and the first 5 principle components of ancestry to control for population stratification in the genetic analysis.

These confounders were not under investigation as hypothesized primary risk factors but were included to examine the effect of confounders on the association of the risk factors under investigation with depression symptoms as well as to mark missing data and participant demographic characteristics.Statistical Analysis

We conducted growth mixture modeling (GMM) in Mplus version 8 (Muthén and Muthén)49 to identify latent trajectories of depression symptoms using 9 measures of the SMFQ. Briefly, GMM stratifies individuals from a population into multiple heterogeneous trajectories (or latent classes).50 

Odds ratios (ORs) and their corresponding 95% CIs were derived from multinomial logistic regressions, where the trajectory with the largest sample size was used as the reference. Multinomial logistic regressions were used to calculate values. Statistical significance was set at P < .05, and all tests were 2-tailed. Further details about model fit and how we assessed the validity of these trajectories can be found in the eAppendix in the Supplement.

Missing data in the GMM were handled using full-information maximum-likelihood estimation.49 We included individuals in the analysis if they had depression symptoms measured on at least 1 occassion.23 Previous research on this data has demonstrated little difference on the shape of trajectories, distribution of trajectory membership, or associations of trajectories with outcomes when comparing individuals with at least 1 measurement of depression symptoms with participants with at least 3 measurements or all 9 measurements.15Results

Data were available for 9394 individuals with at least 1 measurement of depression symptoms. Demographic characteristics for these individuals can be found in eTable 2 in the Supplement, but briefly, individuals included in the study were more likely to be female and have mothers with more education and higher socioeconomic status at birth.

The sample size of individuals with all risk factors and confounders was 3525, including 1771 female participants (50.2%). The mean (SD) age at the first measure of depression symptoms was 10.7 (0.3) years, and the mean (SD) age at the last measure of depression symptoms was 23.8 (0.5) years.

Results from our GMM indicated that a 5-class trajectory solution was best suited the data (eTable 3 in the Supplement). The shapes of these trajectories and class distributions did not differ substantially between the sample of 9394 individuals and the sample of 3525 individuals (eTables 4-6 in the Supplement).Trajectories of Depression Symptoms

Among the sample of 3525 individuals, 5 heterogeneous trajectories of depression symptoms were derived. First, the stable-low trajectory (2506 individuals [71.1%]) included individuals who had consistently low levels of depression symptoms. Second, the early-adult–onset trajectory (393 individuals [11.1%]) included individuals who started with low depression symptoms that increased during adolescence and young adulthood.

Third, the adolescent-limited trajectory (325 individuals [9.2%]) included individuals who experienced elevated levels of depression symptoms only during adolescence, and fourth, the childhood-limited trajectory (203 individuals [5.8%]) included individuals who started with elevated levels of depression symptoms in childhood that decreased. Fifth, the childhood-persistent trajectory (98 individuals [2.8%]) included individuals with moderate levels of depression symptoms that continued to increase and stay high during adolescence and into young adulthood (Figure 1).Association of Risk Factors With Trajectories of Depression Symptoms

Full results from the multivariate analysis of risk factors with varying trajectories are presented in the Table. The risk factors of varying trajectories of depression symptoms are shown in Figure 2. For the following analysis, the OR represents each trajectory compared with the stable-low trajectory. Correlations between risk factors are shown in eTable 7 in the Supplement, and results did not differ in univariate or unadjusted multivariate analyses (eTable 8 and eTable 9 in the Supplement).

Being female was associated with all trajectories (childhood persistent: OR, 6.45; 95% CI, 2.89-14.38; early-adult onset: OR, 1.96; 95% CI, 1.33-2.88; adolescent limited: OR, 6.04; 95% CI, 3.35-10.87; and childhood limited: OR, 1.81; 95% CI, 1.13-2.90). A higher PRS for depression symptoms was associated with the childhood-persistent trajectory (OR, 1.47; 95% CI, 1.10-1.96) and the early-adult–onset trajectory (OR, 1.29; 95% CI, 1.06-1.57) but not the adolescent-limited trajectory (OR, 1.04; 95% CI, 0.85-1.27) or childhood-limited trajectory (OR, 1.01; 95% CI, 0.81-1.25). Maternal postnatal depression was associated with the childhood-persistent trajectory (OR, 2.37; 95% CI, 1.16-4.85) and the early-adult–onset trajectory (OR, 2.39; 95% CI, 1.41-4.07) but not the adolescent-limited trajectory (OR, 1.12; 95% CI, 0.54-2.31) or childhood-limited trajectory (OR, 1.70; 95% CI, 0.80-3.62). Partner cruelty to the mother was associated with the early-adult–onset trajectory (OR, 1.78; 95% CI, 1.05-3.04) and adolescent-limited trajectory (OR, 2.30; 95% CI, 1.36-3.90), but it was not associated with the childhood-persistent trajectory (OR, 1.61; 95% CI, 0.66-3.95) or the childhood-limited trajectory (OR, 1.06; 95% CI, 0.48-2.37). Childhood anxiety was associated with the childhood-persistent trajectory (OR, 1.30; 95% CI, 1.16-1.45), early-adult–onset trajectory (OR, 1.12; 95% CI, 1.01-1.24), and childhood-limited trajectory (OR, 1.23; 95% CI, 1.08-1.41), but it was not associated with the adolescent-limited trajectory (OR, 1.09; 95% CI, 0.98-1.21). Being bullied was also associated with all 4 at-risk trajectories (childhood persistent: OR, 4.91; 95% CI, 2.52-9.58; early-adult onset: OR, 1.73; 95% CI, 1.10-2.70; adolescent limited: OR, 1.56; 95% CI, 1.00-2.44; and childhood limited: OR, 8.08; 95% CI, 4.92-13.26).Discussion

This study identified 5 distinct trajectories of depression symptoms from late childhood to young adulthood. These trajectories were associated with both genetic and environmental risk factors. These findings suggest that examining both genetic and environmental antecedents could help identify groups with severe and chronic depression symptoms (ie, in the childhood-persistent trajectory) who should be prioritized for early intervention. Certain risk profiles also showed specific associations with longitudinal patterns of depression symptoms.

Polygenic risk for depression symptoms (ie, greater genetic liability to depression) was associated with the childhood-persistent trajectory and early-adult–onset trajectory, which supports the notion that genetic liability may play an important role in the onset of depression in adolescence.23,51 

Polygenic risk has previously been associated with both depression and higher trajectories of depression symptoms in later adult populations.52,53 

However, our results suggest that genetic liability to greater depression symptoms may begin to manifest in childhood and adolescence, as previous research has highlighted.23 

As such, genetic liability could be a mechanism for chronic and/or more severe depression symptoms throughout the life course that operates through specific neurological or hormonal systems at certain stages of development.24 

Similar results have been observed for trajectories of attention-deficit/hyperactivity disorder,54 suggesting that genetic liability can affect the development of a trait. However, it is unlikely that genetic liability alone is responsible for the more severe trajectories. Instead, a more plausible explanation is that a complex interplay between genetics and the environment exists.30,31,51 This is not yet understood, and it is also not clear how genetic liability to depression might affect later environmental risk factors.55 Future research should explore this to discover potential pathways and mechanisms involved in the maintenance of depression.

Previous research has shown that bullying is an important factor in the onset of depression in adolescence and adulthood.25,26 However, our findings highlight that being bullied in childhood is associated with both short-term and long-term consequences. Whether exposure to bullying has a lasting effect may depend on genetic liabilities to depression and bullying.2729 

Bullying was assessed at age 10 years (shortly before the first assessment of depression symptoms) and was most associated with the childhood-limited trajectory, which showed severe depression symptoms in early childhood that diminished over time. This could reflect an immediate reaction to bullying that then resolves.

Thus, bullying may be a time-specific factor because it had weaker associations with the early-adult–onset trajectory and adolescent-limited trajectory. However, it was associated with the childhood-persistent trajectory, suggesting there could be long-term consequences for individuals who were bullied in childhood but still have consistently severe depression symptoms more than 10 years later.

The difference between these 2 trajectories could be that the individuals with the childhood-persistent trajectory also have genetic and additional environmental factors that make it harder to recover. This supports previous research highlighting genetic liability to schizophrenia, and subsequent bullying is associated with worse trajectories of mental health.56 

This suggests that individuals who experience bullying who have familial and other accumulating risk factors should be prioritized for intervention. A similar interpretation could exist for childhood anxiety, which showed associations with the childhood-persistent trajectory, early-adult–onset trajectory, and childhood-limited trajectory. Interventions to build resilience early for those with genetic liability may yield the most effective method for the prevention of long-term or severe depression.

Sex differences have consistently been associated with less favorable trajectories of depression symptoms,57 and we observed strong associations of being female with the childhood-persistent trajectory and the adolescent-limited trajectory, supporting previous research.19 

The childhood-persistent trajectory and adolescent-limited trajectory are distinct, yet we may have observed these differences because young women who belong to the childhood-persistent group have a genetic liability to depression in addition to other factors, such as early pubertal timing57,58 and stress reactivity.59

 In contrast, the young women with the adolescent-limited trajectory may not have the genetic liability to depression symptoms and are perhaps only reacting to early pubertal timing or stressful events. However, it is likely that other biological and environmental risk factors underlie less favorable trajectories for girls and young women, and more research is needed to disentangle this association.

Maternal postnatal depression was associated with the childhood-persistent trajectory and early-adult–onset trajectory, likely reflecting the transmission of maternal depression to offspring.6063 

This association likely reflects genetic influences (as these 2 trajectories were associated with the PRS) as well as parental depression possibly affecting brain development in utero.62,64 Alternatively, it could reflect childhood susceptibilities established in infancy and early development that result in later depression.63

Because maternal postnatal depression was not associated with the adolescent-limited trajectory or childhood-limited trajectory (ie, trajectories not associated with genetic liability), it may suggest that maternal depression has more long-lasting associations, possibly through a genetic and environmental interplay. Interestingly, this pattern was not observed for partner cruelty to the mother, which was associated with the early-adult onset and adolescent-limited trajectories but not the childhood-persistent or childhood-limited trajectories (ie, the trajectories associated with bullying).

It could be that partner cruelty operates through a different pathway and does not share the same time-specific effect compared with bullying. Instead, the adolescent-limited trajectory could be reflective of depression symptoms that decrease after adolescence, once young adults are less influenced by family life.

However, partner cruelty to the mother is a rare exposure, and it is possible that the associations of partner cruelty to the mother with the early-adult–onset and adolescent-limited trajectories reflect a lack of statistical power rather than a true association.

University of Michigan
Media Contacts:
Kara Gavin – University of Michigan
Image Source:
The image is in the public domain.

Original Research: Closed access
“Genomic prediction of depression risk and resilience under stress”. Yu Fang, Laura Scott, Peter Song, Margit Burmeister & Srijan Sen.
Nature Human Behavior doi:10.1038/s41562-019-0759-3.


Please enter your comment!
Please enter your name here

Questo sito usa Akismet per ridurre lo spam. Scopri come i tuoi dati vengono elaborati.