Insights Into Suicide and Depression

The article by Docherty and coworkers in this issue (5) provides new genetic leads from a genome-wide association study (GWAS) performed on a relatively large sample of individuals who died by suicide (N=3,413). Although many studies have examined genetic linkages to suicide risk or suicidal behavior, this study is distinguished by its focus on completed suicides. This is important because suicidal ideation is not the same as self-harm behavior, which is not the same as a completed suicide. Results identified 22 genes potentially related to suicide death with a heritability, based on single-nucleotide polymorphisms, of 25%. Perhaps not surprisingly, a number of the implicated genes overlapped with genes associated with schizophrenia and bipolar disorder. Importantly, polygenic risk scores (PRSs) computed from the suicide GWAS data were predictive of suicide in an independent sample. In addition, PRSs for other disorders and traits, including behavioral disinhibition and major depressive disorder, were also predictive of suicide death. Taken together, these data represent an important step forward in understanding how structural genetic variation relates to the risk to complete suicide. The findings point to specific genes that overlap with other major psychiatric illnesses, and although this research is still in its early stages, it suggests the possibility that PRSs could become part of the psychiatrist’s armamentarium for suicide prevention. Dr. David Goldman, chief of the Laboratory of Neurogenetics at the National Institute on Alcohol Abuse and Alcoholism at the National Institutes of Health, provides an editorial on predicting suicide that discusses the importance of the findings in this article and contextualizes the early stage of this research in relation to its potential for eventual clinical use (6).

In this article, using Swedish national samples, Kendler et al. (7) use a unique design to understand factors associated with the parental transmission of suicide attempts and death by suicide. The “expanded adoption design” included comparisons among intact nuclear families, families in which the biological father was not present, families with a stepfather, and adoptive families. Use of this design allowed for an understanding of the relations between parents and children in regard to suicidal behavior for genes only, for rearing only, or for both genes and rearing. Consistent with earlier work, results demonstrated involvement of both genetics and rearing. Interestingly, the parental transmission of suicide attempts was found to be relatively strong compared with the parental transmission of death by suicide. Moreover, the parental transmission to sons for suicide attempts was found to be stronger than it was to daughters. The researchers also found strong effects of parental psychiatric illness on suicide risk that appeared to be mediated by genetic, and not rearing, differences. In addition, the authors draw the conclusion from their data that suicide attempts and death by suicide are not likely to be on a simple continuum but rather may represent the product of different underlying genetics.

It is well known that depression is frequently an initial diagnosis in patients who later are found to have bipolar illness or psychotic disorders. The study in this issue by Musliner et al. (8) examines the extent to which genetic information in the form of PRSs can be helpful in making long-term predictions regarding illness course in patients who present with unipolar depression. Such a capacity would have important implications for treatment planning in relation to psychopharmacologic and psychosocial interventions, as well as for the possibility of developing prevention strategies. The authors examined an extremely large number of individuals from a Danish sample with a primary diagnosis of depression who did not have bipolar or psychotic disorders (N=16,949). Follow-up data over a median of 7 years were used to ascertain outcomes. During this period of time, the conversion rate from depression to bipolar disorder was 7.3%, and the rate for psychotic disorders was 13.8%. Although the PRSs for bipolar disorder and schizophrenia were correlated with each other, when controlling for each of these scores, it was found that only the bipolar PRS was associated with the later progression to bipolar disorder and, likewise, that only the schizophrenia PRS was associated with the later development of psychotic disorders. Interestingly, high PRSs for both schizophrenia and bipolar disorder appeared to be associated with a greater likelihood of developing affective psychotic disorders (psychosis associated with major depression or bipolar disorder). The authors also documented the importance of parental history for both bipolar and psychotic disorders. This effect was particularly notable for bipolar disorder, as individuals who developed bipolar disorder were more than five times more likely to be offspring of parents with bipolar disorder. In contrast, the overall contribution of the PRSs to making long-term predictions about conversion to bipolar disorder or psychotic disorders was small. In his editorial, Dr. Peter Holmans, a professor at Cardiff University and an expert in genetic analyses of traits and disorders, provides a thoughtful presentation on how to think about the value of PRSs (9). In so doing, he highlights their importance in relation to understanding disease phenotypes and in predicting an individual’s degree of risk, and the editorial also sheds light on the molecular mechanisms underlying pathophysiology. He also describes how to assess potential clinical utility, as PRSs are being considered for clinical use.

In this issue of the Journal, Choi et al. (10) use a study design called Mendelian randomization in an attempt to causally link modifiable depression-related risk factors to strategies focused on preventing depression. Using data from the UK Biobank (>100,000 individuals), the authors examined 106 modifiable factors that could be associated with the risk to develop depression. After initially implicating a number of factors that were associated with the development of depression, the authors then used Mendelian randomization to deal with potential confounding effects among these factors, as well as to implicate causality in the relation between an individual factor and the development of depression. Mendelian randomization is a method that is used to take advantage of genetic variation that may be associated with a putative modifiable factor. By stratifying groups based on the genetic variation related to the modifiable factor and demonstrating a relation between the factor-related genetic variation and depression, the inference of causality between the modifiable factor of interest and depression is strengthened. In the initial association analysis, 29 factors, including some of the “usual suspects” (e.g., socialization, sleep, and exercise), were found to be associated with the development of depression. However, the Mendelian randomization method substantiated only a few of these factors as likely to be causally linked—the positive effects of confiding in others, and the negative effects of television use, daytime napping, and, unexpectedly, multivitamin use. This report is accompanied by an editorial by Dr. Joseph Firth of the University of Manchester with coauthors Dr. Robyn Wootton from the University of Bristol and Dr. Andre Carvalho from the University of Toronto (11). The editorial focuses on the importance of developing preventive strategies for impeding the development of depression, explains and discusses the findings in this report, and further emphasizes the value of using the Mendelian randomization approach in studies aimed at linking causal risk factors to psychopathology.

Although as psychiatrists, our professional work is focused on decreasing distress and enhancing our patients’ well-being, we are certainly not immune to the effects of stress. In their article in this issue of the Journal, Summers and coworkers (12) provide survey data from 2,084 North American psychiatrists aimed at understanding work-related factors associated with burnout and depression. The hope is that some of these identified factors could serve as modifiable targets to reduce the likelihood of developing burnout and/or depression. In this article and in his accompanying commentary (3), Summers discusses the characteristics of burnout, describing it as more akin to mental exhaustion. While conveying that burnout shares some features with depression, including negative affect, he emphasizes that burnout is quite different than the syndrome of major depression. The findings from the study are alarming, as 78% of individuals reported burnout scores that were consistent with high levels of burnout, and 16% reported depression scores consistent with major depression. Women appeared to be particularly affected, and female gender and the inability to control one’s schedule were associated with the degree of burnout. Female gender, being early on in one’s career, and practicing in a nonacademic setting were significantly associated with high depression scores. These findings are important as they draw attention to factors that can potentially undermine the well-being of practicing psychiatrists. Future work utilizing prospective designs and following psychiatrists longitudinally will help us more deeply understand the causal factors related to burnout and to depression in the workplace and should be helpful in designing effective mitigation strategies.

Hormonal alterations have long been associated with negative affect, altered cognition, and depression, and numerous studies have examined hormonal treatments as antidepressant strategies. It is important to emphasize that hormones not only affect peripheral physiological processes but also have direct influences on brain function. There is particular interest in the hypothalamic-pituitary-gonadal axis, as sex-related hormones are associated with depressive states, estrogen replacement in postmenopausal women can have positive effects, and a progesterone metabolite, a neurosteroid, has been approved for the treatment of postpartum depression. In men, low levels of testosterone have been associated with depression, and testosterone administration appears to be effective in men with depression who have low levels of testosterone. For more detail regarding the complex relation between hormones and depression and for a review of hormonal treatments for depression, see the review article published in the August 2020 issue of the Journal by Dwyer et al. (13). The article in this issue of the Journal authored by Dichtel et al. (14) reports on findings from a double-blind placebo-controlled study examining the effects of low-dose testosterone cream on augmenting antidepressant responses in women. This well-designed study, inspired by open-label data, is an attempt to formally test the antidepressant value of testosterone in augmenting women’s responses to antidepressants. Specifically, 21- to 75-year-old women who met criteria for major depression and were on a current antidepressant for at least 8 weeks were randomly assigned to receive either placebo or testosterone cream starting at 10 mg/day. Over the ensuing 8 weeks of the trial, individual dosing was adjusted based on free testosterone levels to achieve levels in the upper range of normal. No difference was found between placebo cream and testosterone cream augmentation, as both the placebo and testosterone group similarly responded over the 8-week period, at 47% and 49%, respectively. Additionally, while improvements in fatigue and sexual function were noted, there were no significant differences between the testosterone and placebo groups. The authors also used functional MRI in a small subset of patients in an attempt to understand the neural correlates associated with testosterone’s effects. Not surprisingly, an imaging study with this very small sample did not provide useful insights at the neural circuit level. Although the lack of apparent efficacy reported in this trial should be interpreted in relation to the high placebo response rate, the authors conclude that this approach should “…not be recommended for women with major depression.” Dr. Sanjay Mathew from Baylor College of Medicine contributes an editorial that highlights the strengths and weaknesses of the trial’s design (15). In this context, he discusses the implications of the high placebo response rate observed in this trial and suggests that further work should investigate the therapeutic efficacy of testosterone treatment strategies in women with depression.

Considerable work has demonstrated the importance of the gut microbiome in influencing behavior and health and also in linking the microbiome to psychiatric illnesses. However, the cellular and molecular mechanisms underlying the microbiome linkage to psychopathology remain poorly understood. In an exciting article presented in this issue of the Journal, Medina-Rodriguez et al. (16) report on findings that establish a mechanism by which a specific bacterium in the gut works via the immune system to modulate the expression of depressive-like behaviors. In a number of experiments performed in mice, the authors demonstrate that the bacterium, segmented filamentous bacteria, increases the tendency to engage in depressive-like behavior. This effect appears to be due to the bacterial stimulation of what is termed a quorum-sensing molecule, autoinducer-2, that in turn stimulates the host’s production of serum amyloid protein-1 and serum amyloid protein-2. These molecules have the capacity to enhance the production of a specific immune T cell in the intestine, Th17 cells. The authors then demonstrate that the administration of autoinducer-2 induces depressive behavior and that this effect is mediated by Th17 cells. Furthermore, the inhibition of the intermediary serum amyloid proteins appears to have antidepressant-like effects, and this is associated with a reduction in Th17 cells in the hippocampus. In normal mice, Th17 cells are not found in the brain. Finally, in a small sample of humans, the authors provide preliminary translational evidence for this mechanism. Specifically, they found that individuals with depression had increased fecal levels of segmented filamentous bacteria as well as other downstream components that were found to be important in the induction of the depressive-like behaviors in the mouse experiments. These results are exciting because they provide a clear pathway that links gut microbiome alterations to an immune response, via specific molecular intermediaries, that is associated with depressive-like behaviors. In their editorial, Ms. Joana Cruz-Pereira and Dr. John Cryan, from University College Cork, comment on the significance of these findings and highlight the potential importance of understanding the specific molecules involved in mediating the putative linkage between gut bacteria, immune response, neuroinflammation, and depression (17).

The rich set of papers in this issue of the Journal highlight attempts, with state of the art genetic, molecular, animal model, behavioral, epidemiologic, and trial design methods, to better understand the factors contributing to the risk for suicide and depression. In addition, the commentaries in this issue emphasize the need to support and provide interventions for frontline health care workers during the COVID-19 crisis as well as to be mindful of potential burnout and depression in practicing psychiatrists. Understanding the scientific underpinnings of psychiatric illnesses is critical, and the papers in this issue are a step in that direction. While many of the findings reported here are not ready for clinical use, we should be optimistic about further pursuit of the leads uncovered in these studies. This issue features the breadth of the research that is necessary to fully understand the complexity of the diseases that we deal with, ranging from translational animal model molecular studies, to genetic, familial, and population-based studies, to clinical trials. The future of psychiatry lies in our ability to unravel the molecular, neural circuit, behavioral, and interpersonal mechanisms that underlie psychiatric illnesses. An integration of this information will eventually bring us to a point where we can understand the heterogeneity of the illnesses that we treat and apply this understanding to develop more effective personalized treatments.