Using the Power of a Giant Wisely: Confirming Inflammation in Depression

There are many giants among the broad gamut of population-based studies, but the UK Biobank project definitely is the giant among giants. With more than 500,000 participants recruited from across the United Kingdom, this project has contributed novel knowledge to many medical fields. Although psychiatric phenotypes are not measured in much detail, the UK Biobank’s online mental health questionnaire filled out by approximately 150,000 participants yields relevant information for psychiatry. In this issue of the Journal, Pitharouli and colleagues (1) used UK Biobank data to conduct the largest ever case-controlled study to examine the link between the presence of major depressive disorder and serum levels of C-reactive protein (CRP), a nonspecific marker of low-grade inflammation. The authors’ finding of higher CRP levels in persons with major depressive disorder is in line with those of multiple meta-analyses comparing inflammatory markers such as interleukin-6, CRP, and tumor necrosis factor-α between depressed and nondepressed persons, even at young age (2, 3). In peripheral blood mononuclear cells, gene expression upregulation of inflammatory pathways in depression has also been confirmed (4). All of these findings have boosted the field of immunopsychiatry in the last decade by putting the immune system central in examining interactions between the body, the brain, and the mind.

What makes this study so unique to report on here? Well, it is obviously the superpower of the sample. By using data from 26,894 persons with major depression and 59,001 control subjects, the confidence in a finding is extremely high. In fact, the number of depressed persons in this study is more than 10 times higher than that summarized in recent meta-analyses (2, 3). Such a large sample size also allows for detailed checking on the impact that potential biases or mediating variables have, thereby allowing for better causal interpretation of a relationship. A remaining question is to what extent medication use influences the CRP-depression link, as antidepressants have shown to potentiate inflammatory changes (5). This study illustrates that medication use did not bias findings, as exclusion of antidepressant users did not change any of the results. The authors also explored in detail the extent to which differences in lifestyle, health, early-life trauma, and socioeconomic status played explanatory roles in the link between major depressive disorder and CRP. This is important as these factors have not always been considered in previous studies, though their impact on inflammation is significant. In the study by Pitharouli et al., analyses showed that the effect of body mass index (BMI) and smoking was substantial, as taking these variables into account largely reduced the strength of association between major depression and CRP, although it still remained significant. This illustrates that unhealthy lifestyle has a strong contributing role to low-grade inflammation present in depressed persons. However, it is important to realize that a high BMI does not reflect simply “externally driven” eating habits but also internal metabolic dysregulations due to, for example, genetics, alterations in systems involved in homeostatic adjustments (the hypothalamic-pituitary-adrenal axis, neuroendocrine regulators of energy metabolism, including leptin and insulin, and the microbiome), and brain circuitries integrating homeostatic and mood regulatory responses, all of which are potentially involved in depression (6). In fact, adipocytes produce a wide range of inflammatory markers. Consequently, adjusting for BMI could be argued to be an “overadjustment” for relevant underlying mechanisms, thereby leading to an overly conservative estimation of the “true” association between major depression and CRP.

What about the chicken and the egg question? Obviously, a cross-sectional association between major depressive disorder and CRP does not allow for causal inference. A recent meta-analysis summarizing longitudinal studies confirmed that there is evidence for bidirectionality: inflammation increases the risk of subsequent depression, and depression increases the risk of subsequent inflammation (7). In addition, there is increasing evidence that shared genetics tie depression and inflammation together. The UK Biobank analysis as presented by Pitharouli et al. (1) uniquely illustrates that a genome-wide polygenic risk score (PRS) for depression (based on the largest genome-wide association study [GWAS]) also predicted CRP levels. So, the presence not only of major depression itself but also of a high genetic vulnerability for major depression, not necessarily requiring a history of major depression, is associated with higher CRP levels. In addition, this association was substantially reduced after adjustment for BMI, illustrating that the genetic vulnerability for major depressive disorder is linked to inflammation through pathways that underlie regulation of eating habits or metabolic processes related to obesity. Another genetic study using data from the Psychiatric Genomics Consortium already confirmed the opposite direction of association: persons with a higher genetic vulnerability for obesity and inflammation (by PRSs calculated from a GWAS of BMI and CRP) are more likely to develop depression, especially energy-related symptoms of depression (8). This finding was confirmed in a recent UK Biobank–based Mendelian randomization study, where genetic variants associated with BMI were linked to anhedonia, tiredness, and changes in appetite but not with other depressive symptoms (9). So, the association between depression and CRP is seen not just for the phenotypes themselves but also for their underlying genetic vulnerabilities. This further confirms the causal link between depression and inflammation and their partial origins in a person’s genetic basis.

What are the next questions to be addressed in the field of immunopsychiatry? The generalizability of these findings to more clinical or different ethnic samples needs to be established. The UK Biobank consists of a relatively healthy and wealthy selection of the general population, and identified persons with major depressive disorder reflected lifetime cases, many of whom may not have current symptoms. The link with inflammation could be stronger in more severely depressed clinical samples. Nevertheless, even taking this into account, the overall effect size for the link between CRP and major depression has been shown to be rather small (2, 3). In Pitharouli et al.’s analyses, the standardized beta after full adjustment was only 0.024, which is significant when you have such a large sample size. One could therefore wonder whether such finding has clinical relevance. What it does illustrate is that CRP cannot be used as a biomarker to validly distinguish patients from control subjects. However, it does signal that it is worthwhile to more fully disentangle the underlying mechanisms of immuno-inflammatory processes in depressed patients. CRP is a widely used marker of inflammation. It is easy and relatively inexpensive to measure, its assessment can be done in almost all clinical laboratories, and it is highly sensitive in identifying patients with chronic low-grade inflammation. However, it is not a specific indicator and does not teach us about the many possible underlying routes of molecular immune dysregulation. Although there is evidence that both innate and adaptive immune system dysregulations are linked to depression, it is unknown through which mechanisms of action these affect brain function and symptomatology (10). In addition, the extent to which immune dysregulations in depressed patients are due to genetics, environmental factors (e.g., unhealthy lifestyles), somatic ill health, microbiome dysregulations, or their interactions needs to be further addressed. Addressing these remaining questions will give us more guidance on how to best intervene on immuno-inflammatory dysregulation in clinical practice.

Finally, depression is not a monolithic state; it is a very heterogeneous disorder. Not all depressed persons have levels of relevant inflammation. In fact, in the UK Biobank, in line with findings from other studies, not more than a quarter of persons with major depressive disorder had a CRP level of >3 mg/L. It was not further examined how depressed persons with inflammation compare with those without. There was no examination of the role of discrete symptom profiles, depression history, severity, recency status, or other psychiatric comorbidity. Others have illustrated that inflammation is more strongly present in depressed persons with anhedonia (11) or in those with energy-related atypical symptoms (12). Other research has also confirmed the latter by using genome-wide tools: those with higher genetic vulnerability for inflammation were mainly the depressed persons with such neurovegetative, energy-related symptoms (8). Knowing which patients with depression also have altered inflammatory processes is important, as there is some evidence that standard antidepressant treatments do not work as well in this subgroup (13). In addition, it is this group for whom novel anti-inflammatory treatment may be most beneficial (14). Consequently, more work is needed to understand for whom inflammatory processes play a role in the etiology of depression. The UK Biobank can hopefully address these remaining research questions soon: the work based on the giant among giants is not finished yet.