Clarifying the links between depression, serotonin, and SSRIs
This post is in response to’s post, “If serotonin isn't linked to depression, why do SSRIs even work?” on (requires subscription). These issues require some detailed clarification, are of general interest, and have been coming up in discussions quite a bit over the past year, so I’ll address them here.
Is it true that “serotonin has nothing to do with depression”?
Discussions on the serotonin hypothesis become confusing quickly because there is no precise articulation of what the serotonin hypothesis of depression is, or what the nature of the relationship between various aspects of the serotonergic system and various aspects of depression is supposed to be. As a result, it is not evident that disproving some versions of the serotonin hypothesis means that “serotonin has nothing to do with depression.”
Here is an (incomplete) list of the ways in which we might understand the relationship between depression and serotonin:
Depression is caused by low levels of serotonin in the brain or low serotonergic activity
Depression, generally or in some subset of patients, involves alterations of the serotonin signaling system (e.g. in the distribution or sensitivity of certain sorts of serotonin receptors)
Serotonergic system mechanistically links depressive symptoms and neurobiological dysfunctions in other aspects of brain functioning (e.g. neurogenesis or neuroplasticity)
Serotonin system is generally involved in the regulation of mood and emotions, and there may be no specific abnormality in the serotonin system in depression, by and large, but it still provides us a target for intervention with serotonergic antidepressants.
In addition, we can also talk about whether this involvement of serotonin is considered the central or primary cause of depression or whether it exists as one causal element in a more complex causal web.
We’ve known for decades that the crude version of this hypothesis referring to synaptic serotonin levels is not supported by evidence. Many researchers have believed that some alterations exist in the serotonin system, and some preliminary research findings did suggest that, but nothing conclusive has emerged that commands a strong consensus. The scientific debate is far from over, contrary to Moncrieff et al, 2022. The significance of monoamine depletion remains to be dispelled (in 2007 a meta-analysis in Molecular Psychiatry reported that monoamine depletion produces depressed mood in individuals with a family history of depression and in drug-free patients with depression in remission). A 2015 meta-analysis showed that serotonin transporter availability in depressed patients is reduced in key regions of the limbic system. And even now, just a few months ago in October 2022 and after the Moncrieff paper, the first direct assessment of serotonin release capacity in people with depression reported a reduction in serotonin release capacity in patients experiencing a major depressive episode (it’s a small study and the difference between the groups is not that striking; I consider the findings to be preliminary). There are other studies as well. I am not making a particularly strong claim here, merely saying that the presence and nature of alteration of serotonergic system in depression are open scientific problems.
Aside from the question of serotonin alteration or dysfunction, the involvement of serotonin system in the general regulation of mood and emotions is backed by a large body of literature from animals as well as humans (e.g., see this comprehensive review). Even if there is no dysfunction of serotonin in depression, the link between serotonergic mechanisms and aspects of mood/behavior allows for the possibility of effective intervention. There is nothing wrong with kidneys in chronic heart failure, but we can use diuresis as a treatment; there is nothing wrong with prostaglandin pathways in infections, but we can act on them to treat fever; etc. Serotonergic pathways appear to be mechanistically involved in regulation of mood and emotions. There may not necessarily be anything “wrong,” “dysfunctional,” or “imbalanced” in these pathways (except in an indirect sense) — they may be working just fine — but if they are involved in how mood/emotions are regulated, they can be intervened on to produce desired effects.
I want to make it clear that none of this justifies the common narrative of depression as fundamentally a problem of serotonin. Serotonergic system or even serotonergic alterations may be involved in depression in some complex manner, but depression is a highly heterogenous and multifactorial condition, and involves a range of neurophysiological, psychological, and sociopolitical factors. Given the heterogeneity of depression, it is highly unlikely that such abnormalities of serotonin, even if they exist, will be present in most individuals with depression.
The emphasis on serotonin in depression literature stems from the fact that, for a long time, serotonergic pharmaceuticals produced the most reliable effects on depressed mood among the options available. There is a good reason why one pharmaceutical company after another developed me-too serotonergic antidepressants. If any psychoactive compound could do the job and demonstrate efficacy in phase 3 trials, serotonergic antidepressants would not have dominated the market for 3 decades.
Another point to consider: neuroscientists understand causality in different ways. A somewhat non-traditional but scientifically well-respected understanding of causality is the notion of “causal dependence” according to which if an outcome depends on the existence of a mechanism and that mechanism can be targeted by an intervention, then the mechanism may be conceptualized as a cause, even if there is nothing abnormal about the mechanism itself. The advantage of this approach is that it provides ways of identifying therapeutic targets. The sort of dependence appears relevant to serotonin and depression.
On the efficacy of antidepressants
I have written in detail about this issue previously:
SSRIs show consistent superiority over placebo in reducing depressed mood, and a network analysis published earlier this month in Molecular Psychiatry reported that SSRIs have quick and strong direct effects on depressed mood and psychic anxiety, with indirect effects on cognitive symptoms.
To point to additional ways in which serotonergic interventions could produce desired effects on mood and behavior, here are Carhart-Harris and Nutt from a 2017 paper on serotonin and brain function:
“… serotonin neurotransmission enhances two distinct adaptive responses to adversity, mediated in large part by its two most prevalent and researched brain receptors: the 5-HT1A and 5-HT2A receptors. We propose that passive coping (i.e. tolerating a source of stress) is mediated by postsynaptic 5-HT1AR signalling and characterised by stress moderation. Conversely, we argue that active coping (i.e. actively addressing a source of stress) is mediated by 5-HT2AR signalling and characterised by enhanced plasticity (defined as capacity for change). We propose that 5-HT1AR-mediated stress moderation may be the brain's default response to adversity but that an improved ability to change one's situation and/or relationship to it via 5-HT2AR-mediated plasticity may also be important - and increasingly so as the level of adversity reaches a critical point. We propose that the 5-HT1AR pathway is enhanced by conventional 5-HT reuptake blocking antidepressants such as the selective serotonin reuptake inhibitors (SSRIs), whereas the 5-HT2AR pathway is enhanced by 5-HT2AR-agonist psychedelics.”
There may also be other ways in which antidepressants produce effects, with or without serotonergic mediation, such as mechanisms of neuroplasticity.
Medieval trepanations for mental illness
[Added clarification: I mean that there is no solid historical evidence that trepanation was used to treat madness or mental illness in medieval times. Trepanation was definitely used for other medical and neurological ailments.]
Psychiatric Drugs as Mere Perturbations model (PDMP)
Hoel offers an intentionally simplistic toy hypothesis:
“… with monumental papering over and coarse-graining, we can think of depression as neural dynamics being “in a rut” wherein constant anxiety and negative thoughts stack on top of one another. Let’s forget precisely why neural dynamics would get into this rut. But if psychiatric drugs work via randomly perturbing the brain, then change in any direction at all is sometimes enough to get the brain away from its abnormal ruts and toward a more normal state wherein cognitive homeostasis can take back over. Note that we don’t have to understand anything about the mechanism of action here, nor does the perturbation have to have any specific effect on the cause of the disease. We know that drugs can perturb brain activity, we essentially give drugs at random until we find one that shakes the brain far enough away in the right direction from whatever the underlying problem is.”
Hoel is using this to point towards a complex possibility. He makes it clear: “I’m actually not suggesting that the PDMP is really true.” So I am not going to take the toy hypothesis and criticize it literally, but I’ll make some points:
Randomly perturbing the brain doesn’t seem to work for depression or in fact for any psychiatric disorder. There is a wide range of ways in which the brain can be perturbed by psychiatric medications to produce therapeutic improvement, but the number of ways is still limited, and it isn’t random. Some perturbations produce therapeutic effects more consistency and reliably than others.
The perturbations may not be directly linked to the cause of the disease, but they have to be mechanistically related to symptom production and maintenance. If we perturb brain activity in an area or network that has little to do with mood, we will see little to no effects.
If we conceptualize depression as a “symptom network,” a cluster of symptoms that has become self-sustaining due to direct causal connections between symptoms, we can think of a variety of ways in which brain functioning can be altered or perturbed to disrupt the self-sustaining feedback loop, but again these ways will neither be random nor mechanistically unrelated to the neural dynamics of depression.
Psychiatry at the Margins is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
Thanks for writing this up, it was an interesting read! A couple points:
(a) Are you sure that electroshock therapy could not be described as a "random" perturbation? (with randomness, as I said originally, not meaning statistical white noise, just like untargeted perturbations that don't have anything to do with the etiology)
(b) Are you sure that trepanations were indeed never performed in medieval times, only fantasized about? Here's from Wikipedia (which may be wrong about this, but it seems pretty confident and links sources) https://en.wikipedia.org/wiki/Trepanning
"Hippocrates gave specific directions on the procedure from its evolution through the Greek age, and Galen also elaborates on the procedure. During the Middle Ages and the Renaissance, trepanation was practiced as a cure for various ailments, including seizures and skull fractures. Out of eight skulls with trepanations from the 6th to 8th centuries found in southwestern Germany, seven skulls show clear evidence of healing and survival after trepanation, suggesting that the survival rate of the operations was high and the infection rate was low. . . .
During the 16th and 17th centuries, around 80% of people survived the procedure of trepanation."
I suppose one could lean on the notion that "various ailments" don't include any mental disorders, but I'd be interested in more about the history of it.
Thanks for making this critical point, Awais:
"Serotonergic system or even serotonergic alterations may be involved in depression in some complex manner, but depression is a highly heterogenous and multifactorial condition, and involves a range of neurophysiological, psychological, and sociopolitical factors."
The general public's perennial fixation on serotonin--exacerbated by Big Pharma ads and the misleading and muddled "umbrella review" by Moncrieff et al--is partly a consequence of inadequate "messaging" by professional psychiatric organizations and academic psychiatry. All of us (including this writer) could have done a better job in communicating to the public the vast heterogeneity of "depression" and the immense complexity of neurobiological, psychological and sociocultural factors in its genesis. And all of this is aside from the well-established, albeit modest, efficacy of antidepressants in treating at least a significant subgroup of patients with major depression.
Not incidentally, it is worth noting that the whole notion of "serotonergic" antidepressants is misleading to begin with. Sertraline, for example, also affects reuptake of dopamine, while paroxetine has effects on norepinephrine. [see, e.g., Richelson E, J Clin Psychiatry 64 (suppl 13)5-12, 2003). Finally, of course, the antidepressant bupropion [Wellbutrin] has almost no effect on serotonin--it enhances noradrenergic and dopaminergic function--and yet is an effective agent in depression. All this has been known for many years, but was never fully explained to clinicians or the general public.
Most likely--and somewhat ironically--these neurotransmitter effects are largely a "sideshow" vis-a-vis the proximate mechanism of action of most antidepressants, which probably involves enhanced expression of neurotrophic factors like BDNF and enhanced "signalling" among neural networks.
Dr. George Dawson and I explore these issues in detail in Psychiatric Times, and companion pieces by Dr. Alexander Lisinski and me explore the issue of antidepressant efficacy. The links follow, FYI.
Ronald W. Pies, MD