Elias Eriksson is professor at the Institute of Neuroscience and Physiology at the University of Gothenburg, Sweden. In recognition of his research contributions to the pharmacology of serotonin reuptake inhibitors, he is this year’s winner of the ECNP Neuropsychopharmacology Award.
He will deliver his Award Plenary Lecture on ‘The efficacy and mechanisms of action of SSRIs – lessons learnt from premenstrual dysphoric disorder and depression’ at the 37th ECNP Congress in Milan, Italy, on 21-24 September 2024.
Professor Eriksson recently spoke to the ECNP Press Officer, Tom Parkhill.
Professor Eriksson, you have been awarded the 2024 ECNP Neuropsychopharmacology Award, let me offer you congratulations on that.
Thank you.
Tell me a little bit about yourself, where are you from, and how did you come to work in science?
I'm from Gothenburg, on the west coast of Sweden. I started to study medicine right after school, with plans to be an MD and maybe a psychiatrist. During my MD programme I started PhD training in the Department of Pharmacology, which at that time was led by the famous Arvid Carlsson, who won the Nobel Prize in 2000, and who many view as the pioneer of the entire field of psychopharmacology. He was, for example, the one who discovered that dopamine is a transmitter modulating motor function, leading to a treatment of Parkinson’s disease, and also the first to suggest that antipsychotics block dopamine receptors.
So I started there, and I liked it so much that I never left. I have been in the same department for my entire career, so maybe not a very thrilling and exciting career! This is the department of Basic Pharmacology, but I have also done a lot of clinical research, so I have worked partly in clinical and partly in preclinical research.
You've been involved in several different fields. You started working in the field of panic.
We have studied different psychiatric disorders, but with underlying similarities, and mostly those responding to serotonergic drugs. We have also worked on dopamine and schizophrenia – we were, for example, the first to report that certain antipsychotics are inverse agonists rather than antagonists at dopaminergic D2 receptors. But our main focus has been on the SSRIs. And our first clinical studies dealt, as you say, with panic disorder.
This was in an era just before the SSRIs were introduced, but there was a very SSRI-like compound called clomipramine that was one of the old tricyclic drugs. Almost 30 years earlier the US psychiatrist Donald Klein had shown that another tricyclic drug, imipramine, was effective for panic attacks, but this observation had been largely neglected by clinicians, and no drug was approved for this indication. My colleague Kjell Modigh, who was a psychiatrist, however, had read the papers by Klein, and had the impression from his clinical work that patients responded markedly better to clomipramine than to imipramine. We found that intriguing because clomipramine acts mainly on serotonin and at that time most people believed that panic was related to noradrenalin, not serotonin. We were only few who advocated the idea that serotonin could be important for the anti-panic efficacy of tricylics.
So my entry into clinical research was a placebo-controlled study where we could confirm that clomipramine is indeed better than imipramine, supporting our theory that it is serotonin you should target in panic. Our activities led to clomipramine being the first antidepressant ever to be approved for this indication. It is highly effective for panic at low doses and became widely prescribed in Europe. And then came the SSRIs, which were also found to be effective for panic and largely replaced clomipramine because of fewer side-effects. So that was how we became interested in the non-depressive indications for this class of drugs.
So this was just before the SSRI's came out, which I guess would have been Prozac, in the mid 1980s.
If you had asked Arvid Carlsson he would have said “no, no, no, Prozac was not the first”. The first was an SSRI that Arvid Carlsson had developed with Astra Pharmaceuticals in Sweden called zimelidine. It was shown to be effective for depression and clinically introduced but unfortunately withdrawn because of side-effects. And then came Prozac and the other SSRIs. So Prozac was the first blockbuster SSRI but not the first to be patented as an antidepressant.
This led you more generally into working with serotonin. Where did you go from there?
Our panic study made us intrigued by the fact that one group of drugs is effective for very different conditions, such as panic and depression – and OCD, for which clomipramine (and later the SSRIs) was also known to be effective. This was however before SSRIs were shown to be helpful also in generalised anxiety disorder and social phobia. But quite by chance our next field of interest was not an anxiety disorder but premenstrual dysphoria.
How did you get into working with premenstrual dysphoria? That’s a more physiological area?
Yes, I think it is physiological in the sense that premenstrual complaints are probably a normally distributed physiological condition in the female population rather than something aberrant. We had been studying the influence of sex steroids on serotonin in animals – for example observing differences between male and female rats with respect to serotonin turnover – and had hypothesised that a major role for brain serotonin is to modulate sex steroid-driven behaviour. I was approached by a gynaecologist who had written his thesis about premenstrual syndrome but didn't know much about the brain, and he knew that we were interested in that field. He said there was still no treatment for this condition and asked us if we could suggest anything. We tested clomipramine, the same compound we had used for panic, and it had indeed a remarkable effect. This was, I believe, the first report on serotonin reuptake inhibition for this condition, but at almost the same time there was a study suggesting fluoxetine to be effective.
Impressed by the marked efficacy of clomipramine for this condition, we started to do a lot of work in PMS, which is a very interesting condition for two reasons. Firstly, because of the obvious involvement of sex steroids oestrogen and progesterone in triggering the symptoms, and secondly because the effect of clomipramine and the SSRIs is so impressive; there’s no other indication where the SSRIs are as effective as in premenstrual dysphoria. With respect to the cardinal symptom, irritability, we found that more than 80% of treated women report at least an 80% symptom reduction. One often calculates effect sizes versus placebo for different conditions and for different drugs; a score of around 0.3 is regarded as a low effect size but still an effect, 0.5 is regarded as a moderate effect size, and 0.8 as a large effect. In most conditions in psychiatry, we are around 0.5 or so, but with respect to reducing self-rated premenstrual irritability, the effect size for SSRIs is 1.5.
Also, unlike depression, unlike panic and other anxiety disorders, and unlike obsessive compulsive disorder, we noted that the SSRIs display a rapid, almost immediate, onset of action in PMS. That prompted us to advocate so-called intermittent SSRI treatment; the women only take the medication for one or two weeks prior to menstruation, then they can stop because their symptoms go away.
We also believed that the marked difference between the different indications with respect to onset of action – in panic you often even see a marked aggravation of anxiety during the first days of treatment – raises interesting questions about the underlying mechanisms of action. I don't claim that we have any definite answers to these, but there are some reasons to suggest that the reduction in premenstrual irritability, and other SSRI effects with a short onset of action, are the result of enhanced extracellular levels of serotonin, whereas the reduction in anxiety upon long-term administration may, at least partly, be due to a down-regulation of an increased, anxiety-promoting impact of serotonin.
It's very interesting to hear about how you've been working with gynaecologists because sometimes in medicine people tend to be a little bit compartmentalised. You know, they tend not to think about working with other with other groups.
Yes, you're absolutely right. The fact that clomipramine had not been shown effective in PMS prior to our study can only be explained by the fact that these conditions have been treated primarily by gynaecologists and therefore not of any interest to psychiatrists. This was my first reaction when we had noted the dramatic effect of clomipramine – “has really no one tested this before?” There can sometimes be an unfortunate lack of communication between disciplines.
I work for the ECNP, but I also have done some work for the different groups in the past, including working with a lot of gynaecologists. It's interesting to see how they can look at problems in a completely different way. I think that's sometimes difficult to overcome.
Absolutely, completely different. And it's very true with premenstrual syndrome. Gynaecologists have always regarded all possible premenstrual complaints, including physical symptoms such as headache and breast tenderness, as parts of one and the same syndrome. While psychiatrists are of course mostly interested in the mood symptoms, some arguing that premenstrual dysphoria is a different condition than PMS dominated by physical complaints.
I see that you and your collaborators have managed to rebut many of the recent claims, questioning the usefulness of SSRIs for treatment of depression. I don't want to say we're at the level of conspiracy theories, but a lot of people genuinely don't believe that SSRI's perform any useful function. What are your thoughts on this?
There are conspiracy theories out there. Some critics claim that clinicians, researchers and authorities in different countries – and of course the drug companies – are in some kind of conspiracy, knowing that these drugs are ineffective and harmful, and yet advocating their use. It's quite similar to the anti-vax movement but targeting psychopharmacology in general and antidepressants in particular. So you are quite right. But there are also some more serious academic scientists who have expressed doubts regarding the usefulness of antidepressants.
Here our studies have dealt mainly with depression and started 10 years ago when I read a book by an American psychology professor named Irving Kirsch, who claimed that antidepressants were not effective at all. He had become very famous for this book and for related papers in the scientific literature based on trial-level meta-analyses.
For me it is obvious that antidepressants are effective. They are far from perfect drugs, and not effective for everyone, but even my short experience as a clinician made it obvious to me that these drugs are effective and sometimes miraculously so. So Kirsch must be wrong, I thought. I was invited to give some lectures where I presented data from the literature that I believed rebutted the claims of him and others, but I also wanted to look into why the outcome of the placebo-controlled SSRI trials, admittedly, has often been quite lacklustre.
Irving Kirsch had based his arguments on meta-analysis where he had analysed the outcome of several trials where mostly SSRIs had been compared to placebo. These were traditional meta-analyses where you look at the outcome on group level in different studies. What he claimed was, firstly, that there is no dose response relationship with these drugs; secondly, even if they are superior to placebo with respect to statistical significance, the actual difference between groups is too small to be clinically meaningful; and thirdly, that if they have any effect at all, which he doubted, it was only in those with very severe depression, not in moderate cases. He also said that if they have an effect, it is probably due to the side-effects; when patients experience side-effect, they realise that they have been allotted to the active treatment group and then they experience a reinforced psychological placebo effect.
I thought that these points needed to be addressed. It was well known that many of the early trials comparing SSRI with placebo didn't show any difference – half of them did and half of them did not – which indeed is a bit discouraging. So we thought that we must look into this further, doing our own meta-analyses, but not at trial level, but at a patient level, looking at data from individual patients. This was 10 years ago or so, when such analyses were less common than they are today.
I was lucky enough to recruit a PhD student, Frederik Hieronymous, who is much more talented in mathematics and handling of large databases than I am, and who – together with other talented PhD students – has been essential for the undertaking of this project. We approached the companies that had manufactured the four major early SSRIs: Lundbeck (citalopram), GlaxoSmithKline (paroxetine), Pfizer (sertraline), and Lilly (fluoxetine). At first, most of them said no, you cannot have patient-level data, we have never given out patient-level data. I however had some connection with Lundbeck because I had done a PMS study with them and because Copenhagen is close to Gothenburg. So I contacted some people there and they said OK. That got the ball rolling and led on to us getting data also from Pfizer and Glaxo Smith Kline. Eli Lilly didn't have the very early studies on Prozac in electronic formats, so they couldn't provide that, but they did provide data on their SNRI, duloxetine, so we got that instead. We now had patient-level data for all relevant early studies with three SSRIs and one SNRI. And of course, we checked with the FDA and EMA to make sure that we had received all studies, not just the successful ones, which are those usually being published.
We think that our subsequent analyses, and the work of others, have rebutted many of the common arguments against SSRIs. Perhaps the most important aspect in terms of methodology is that most studies, and indeed the meta-analyses by Irving Kirsh and others, were based on the old Hamilton Depression Rating Scale, introduced by Max Hamilton in the 50s. Because of conservatism, most companies had used the same scale, even if they probably realised that it was far from optimal. The problem with this scale is that it covers 17 items, many of which are often rated zero at baseline – since you cannot get any better than zero, including such items only enhances variation but not the possibility to detect a response. But even more important is that many of the symptoms are complaints that you may experience even if you're not depressed, such as gastrointestinal problems, back pain, difficulty in sleeping and so on. So even if you study a normal non-depressed population, they will not score zero on the Hamilton scale. And the most important problem when it comes to using the Hamilton scale in SSRI trials is that it captures common SSRI side-effects as if they were depression symptoms. SSRIs thus can lead to a reduction in weight, and that causes points on the Hamilton. And they can disturb sleep, they can give gastrointestinal symptoms, they can reduce libido, which are all symptoms captured by this scale. This means that you can be in complete remission from your depression, but have some of these side-effects, and then you may be a non-responder according this scale.
Other groups had since long raised these issues, but we wanted to study them in depth by conducting item-based analyses rather than relying on the conventional effect parameter, i.e. the sum score of the 17 items. Other groups had previously constructed various subscales, with different core depression symptoms, but if you do that, critics will say that you have cherry-picked certain symptoms to support your case. So we decided to focus on just one symptom, i.e. the first in Hamilton’s list – depressed mood. Everyone would agree that depressed mood is a core feature of depression, so a treatment effectively reducing depressed mood cannot be without value. When we looked at the traditional effect parameter, the sum score on the Hamilton scale, only 44% of all comparisons of SSRI versus placebo showed the SSRI to be better. But when we instead used depressed mood as effect parameter, SSRIs outperformed placebo in 91% of the trials. In subsequent studies, we could also show that there is indeed a dose response association for the effect of SSRIs, and that the effect size for the reduction in depressed mood in trials after exclusion of suboptimal doses is pretty decent: 0.5 to 0.6. This is on par with most treatments in somatic medicine, so it's nothing to be ashamed of.
In one study we addressed the issue of severity and were able to show that SSRIs were in fact as effective in moderate as in severe depression. The belief that they were only effective in severe depression was found to be a statistical artefact. And then we also wanted to explore the theory that it's the side-effect that causes the response. So we excluded subjects not reporting any side-effects, but still saw a clearcut difference between placebo and the tested SSRI. So yes, we think that we have provided quite conclusive data that these drugs are certainly not effective for all, but definitely superior to placebo.
And for the future?
Our own research is presently focused on SSRI-resistant depression, where we believe that compounds impacting dopaminergic transmission in a novel way may prove efficacious. With respect to psychiatry in general, one should of course rejoice that there has been a remarkable progress with respect to methodology in recent years, enabling a much more sophisticated analyses of brain function in both animals and man than previously. At the same time, it is worth emphasising that we, due to the complexity of the brain, still know very little about the biology of psychiatric disorders, and that it may take rather long time before we have sufficient insight to design optimal drugs. Hence, there is still a place for the traditional form of psychopharmacology, where progress is made by modulating the profile of existing drugs (once discovered by serendipity), and by testing old drugs, or drug combinations, for new indications. Psychiatry may take pride in the fact that pharmacological treatment of psychiatric disorders often is very effective – more effective than often claimed in media – but one should also admit that our insight into why the drugs are effective is still very limited.
Professor Eriksson, thank you very much.
Thank you!
More information on the Plenary Lecture, and biographical details for Professor Eriksson, can be found here.
Read the press release about Elias Eriksson winning the 2024 ECNP Neuropsychopharmacology Award.
Sunday, 22 September, 16.40-17.25, Gold Plenary
PL02 – ECNP Neuropsychopharmacology Award Lecture – The efficacy and mechanisms of action of SSRIs – lessons learnt from premenstrual dysphoric disorder and depression
Register for the 37th ECNP Congress via the button below:
![ECNP register here](/-/media/Images/ecnp/CallToActionButtons/ecnp-register-here.jpg?h=35&w=200&hash=8F1942AD959FAC54DE272B4DA45F01B91591D0EC&la=en)
|
|
|