A series of interviews with Russian scientists involved in a JAMA Psychiatry article

Olga Karpenko, Deputy Editor-in-Chief of Consortium Psychiatricum, Candidate of Medical Sciences, psychiatrist

Philipp Khaitovich, Professor at the Skolkovo Institute of Science and Technology, Head of the Comparative Biology Group of the Institute of Comparative Biology in Shanghai

Elena Stekolshchikova, Candidate of Technical Sciences, Senior Researcher at the Center for Neurobiology and Neurorehabilitation, Skoltech

Anna Tkachev, Researcher at the Center for Neurobiology and Neurorehabilitation, Skoltech

O.K: Good afternoon. Today we are visiting Philipp Khaitovich’s research group at the Skolkovo Center. These are the authors and, I think, the inspirers of the article recently published in the journal Jama Psychiatry. The study was devoted to lipid profiles in patients with severe psychiatric disorders. Hello, Philipp.

P.Kh: Hello.

OK: Will you introduce your colleagues to us? 

P.Kh: First, I would like to introduce this place. This is not exactly Skolkovo, this is the Skolkovo Institute of Science and Technology, because Skolkovo is a whole campus, an innovation center. And here we have the university and our laboratory of molecular neurobiology. And now I will introduce my colleagues: Senior Research fellow Elena Stekolshchikova and Research fellow Anna Tkachev. We all have different academic backgrounds. Lena is a chemist, Anna is a mathematician, and I am a biologist.

O.K: And I am a psychiatrist. And I must say, I was struck by the fact that a group of Russian authors managed to get published in such a journal as JAMA Psychiatry. I may be wrong, but it doesn't seem to have happened in the last ten years. 

P.Kh: This sounds a bit unexpected to me.

O.K: And our colleagues from abroad publish there more or less regularly. But for them, too, it is usually an achievement, an event. Therefore, I congratulate you all. I want to talk about the essence of your study. For me, as a psychiatrist, this is surprising. In psychiatry, I am used to interacting with large entities. For example, with clinical manifestations of psychiatric disorders. And you have entered the microcosm and obtained interesting results. I suggest that we talk about them and, in general, about how to get published in such a prestigious journal, how to organize such a study. First of all, I want to ask you this. You took patients from different centers: in China, in Germany, in Russia. It turns out that scientists around the world are studying lipids? So this is a common research subject for neuroscience? Or how did it happen?  

E.S: I would say that lipidomics as an “omics” science has been actively developing only in the last 10-20 years, which is directly related to the development of the analytical method that we used, chromatography-mass spectrometry. This method allowed the analysis of the lipidome in all patients who participated in our study. In the past, investigators used to focus on other types of molecules. For example, on the composition of the genome, on how pronounced the patient’s RNA expression is: for example, the protein composition of various tissues or blood plasma (which we studied). When the method that we used began to develop, and an opportunity arose to find out exactly the composition of the lipidome in a short time, rather than looking at something than had been investigated previously, it started to be used more widely in various fields. Not only in neuroscience, but also in other clinical areas.

O.K: And how did you proceed? Did you request samples? Or did you initially agree with everyone that you would conduct such a study, collect and analyze blood samples using this procedure in Germany, in China, and in Russia. How to organize such a thing? 

PK.Kh: In many ways, we were lucky because people showed an interest. We are not clinicians, and in this study, we completely rely on samples provided by healthcare organizations and psychiatrists. As Lena correctly stated, this procedure was not used so actively before, and lately it is really becoming more and more common. When we proposed the use of this procedure, we had a good response from groups both in Germany and in China. In fact, they have been collecting these samples for a long time, but initially, of course, they planned to use them for genetic studies, for proteins, for RNA, etc. Why did we start investigating lipids? Because if DNA analysis gave unambiguous markers, then why would you need to look at everything else? Everything would be clear without it. At the very least, everything would be much clearer. Unfortunately, other methods did not help create tests or diagnostic procedures, explain mechanisms of diseases, so we just started to use additional technologies. Lipid analysis just turned out to be such an additional technology.

O.K: So this technology is your invention. 

P.Kh: Of course not.

A.T: Rather, it was our initiative. There was an interest in lipids in psychiatry. It seems to me that we were the catalyst for the study of lipids.  

OK: How did you come up with the idea in the first place? I understand why DNA, RNA, proteins would have to be studied. But why lipids? What is the connection between lipids and psychiatric disorders?

P.Kh: Historically, of course, everything starts with the brain. And we, like many other groups, primarily study the brain because it is definitely important to understand the way the disease affects it. First, we looked at the brain under a microscope, sequenced, performed histological examinations.

OK: Freud also did this in his time.

P.Kh: Not just him. This is a procedure that dates back more than a hundred years. But, as you can see, it did not give an unambiguous result. We have other methods now. For example, structural MRI that helps evaluate the genome and the activity of genes in the cells. But when we looked at the lipid composition of the brain, for example, in schizophrenia, we saw quite remarkable changes there. Quite remarkable compared with the gene activity, with other parameters that we and other groups have measured. The lipid changes were simply more pronounced, so it became obvious that it made sense to determine them in the blood plasma as well. 

O.K: And how do lipids get from the brain into the circulating blood? How do you even understand that it is them, and not some other lipids?

A.T: Normally, lipids from the brain should not enter the blood. If they do, this signals some very serious disorders, such as severe brain injury.

P.Kh: Or complete destruction of the blood-brain barrier. At very advanced stages. That’s not to say that brain lipids directly entered the plasma. This also happens, but this is not our case. 

O.K: Well, then, how do you make a link between the lipids circulating in the plasma and the lipid changes that you have observed on brain microscopy? 

P.Kh: We performed mass spectrometry. How to make a link? I don’t know. We do not have a model at the moment. Is it possible to build such a predictive model? Model the plasma and brain metabolism so that we can establish and predict this relationship? We, in our lab, do not know how to do this. I know of labs that do this. But there are still so many insufficiently studied parameters that the model can practically yield any result. Therefore, we are currently relying on empirical data. In fact, when we started this study, about five years ago, there were already some fragmented studies in this area. With a small number of patients and control groups. They showed certain changes in the lipid composition of the plasma in schizophrenia, in depression, in bipolar disorder, but mostly they were small studies with small cohorts and limited coverage of lipids, the blood lipidome. We had an understanding that some changes were possible there, so it was not quite a wild shot. But at that time there were no studies investigating this issue systematically, and we decided to do a systematic study. Why did we take different cohorts? To assess the reproducibility. In articles that appear in good journals, we see exactly this approach: to take not just one cohort of patients, even if it’s thoroughly selected and very well balanced, containing 100-200 people, but heterogeneous cohorts from different locations, perhaps from different hospitals, even with different inclusion criteria. And to check the results obtained in a cohort in other cohorts. Now this is becoming quite common practice.

O.K: So you have identified a certain fact: in severe mental disorders - such as schizophrenia, bipolar disorder, and depression - there are some common patterns in lipid changes. And we do not yet know exactly how much they relate to what is going on in the brain. How accurate is this diagnostic tool? Can you take a blood sample and determine from the lipid profile that a person, for example, is most likely to have schizophrenia?

P.Kh: We performed tests on the cohorts provided by the Alekseev Clinic. We did not know the diagnosis and Anya calculated the diagnosis from the model.

A.T: Blindly.

O.K: So you were given someone’s blood samples, presumably from patients…

P.Kh: It was not only patients, that was the point. There were several different diagnoses and a control. But we did not know the proportion, nor the identity of the samples.

O.K: That is, there were healthy people and patients with all kinds of diagnoses.

A.T: Schizotypal personality disorder, first episode of psychosis.

O.K: And what is the accuracy of your estimates? 

P.Kh: For schizophrenia, the accuracy was high. Controls and patients with the first episode of classical schizophrenia were very well differentiated. There were about 60 controls, and two of them appeared strange. Based on the model’s estimates, it was more likely that they had a diagnosis. The same with the schizophrenia group: there were about 50-60 people in it. And there were about 5-6 people whose results looked like those of the control group. That is, the accuracy is about 90%. 

O.K: Can you diagnose schizophrenia based on lipids with a ninety percent probability?

A.T: You have to keep in mind that this is a conditional probability. The cohort was balanced, with equal numbers of mental patients and controls. In this case, the accuracy was 90%. But still, in the general population, the probability of severe psychiatric disorders is very low. Therefore, such a test does not allow us to be sure about the diagnosis. To make the diagnosis, we need additional studies. There may be some factors other than psychiatric disorders that can affect the blood lipidome.

O.K: I think 90% is still a very good result. 

P.Kh: That’s why we got our article published in a good journal. If we had had 50%, then, of course, no one would have accepted the article. 

O.K: For clinical psychiatry, 70% is already a very good result.

P.Kh: The point is that the lipids performed much better than anyone had expected. I remember showing these results to my colleagues in Germany, at the Ludwig Maximilian University of Munich, the cradle of classical psychiatry. They were very surprised, because they had previously worked with the genome, with proteins, with many other molecules, with circulating RNAs, and had tried many different methods. Nevertheless, the results we showed them were incomparably better than what they had seen with other types of molecules. However, as Anna quite correctly said, here we enter a territory that is not as well known as, for example, the genome. We still have to study thousands and thousands of samples, thousands and thousands of patients, and controls, and people with different diagnoses, with different symptoms, having different treatments, to finally understand the relationship between lipids and blood plasma. To understand what is going on not only in the brain, but also in human behavior and how the patient responds to treatment. Can we detect these patterns? What, after all, does all of this show? Now, we simply do not have enough data. We cannot say, for example, that a change in the plasma concentration of a particular group of lipids or a particular lipid means that the patient will have more pronounced symptoms or a worse response to treatment. We do not have such evidence yet.

O.K: I think that this is still the beginning of a new era in psychiatry, in medicine, in lipidomics. Before that, there was the genetic era, and now we are moving on to another level.

P.Kh: I would not say this is the beginning of another era, to be honest. First, genetic research is ongoing. I do not think it has completely exhausted itself. Of course, a lot has already been done. Unfortunately, this did not produce the result that everyone had been hoping for: some genetic test for a predisposition or risk of a disease. If this could be done, then nothing else would make any sense. Now different laboratories are testing different types of molecules, including lipids. Why not? The brain is mostly made up of lipids, after all.

O.K: Perhaps it is too early to ask this question, since no coherent hypothesis has been fully formed yet. But what do you think is the primary cause: changes in brain lipids that eventually lead to a mental disorder, or something that happens during the development of the mental disorder and changes the brain?

A.T: We can only speculate. But if we have to answer such a question, it seems to me that these are strongly connected things. And it doesn’t even have to be brain lipids or plasma lipids. As far as I know from clinical studies, people with psychiatric disorders generally have a large number of health problems. They can have, for example, cardiovascular problems and so on. I think mental health is in some way related to the physical condition of the body; they cannot be separated from each other. Probably, the lipid profile that we see is a characteristic part of the body condition in these people, and is a side effect of the development of the disease. At least, I have such a feeling. 

E.S: I agree. The question is not so much if the brain lipids have changed and somehow entered the blood, it’s more about the fact that there is probably some regulation impairment. The blood plasma reflects a huge number of processes going on in the body, and we are able to detect these changes in all three cohorts. As Anna rightly said, these are very complex interconnected processes that we can observe.

O.K: Another interesting thing. Now we are on the verge of creating a new classification of psychiatric disorders, ICD-11. There is an ongoing debate around the criteria for schizophrenia, for example. And many say that cultural features should be taken into account. In each chapter, there is always a separate section about the cultural characteristics of the manifestations of the same disorders. You have taken three large samples: a Chinese, a Russian, and a German. As I understand, they had comparable lipid profiles. They had a diagnosis of schizophrenia, bipolar disorder, or depression, and you evaluated the comparability of the lipids. And they turned out to be the same? 

P.Kh: No, not the same.

A.T: They turned out to be different, and this was the idea of the whole study, a key part of it. The fact that we have three different cohorts and that they are so heterogeneous. This is not a limitation of the study, but, on the contrary, a key factor that allows us to make conclusions that are more reliable. Of course, different countries and cultures may have slightly different diagnostic criteria for psychiatric disorders. In addition, there may be many other cultural factors that make these data sets very different. For example, nutrition or how people with mental disorders live. These factors are very hard to take into account directly, they cannot be written out in tables, but they can be reflected in the blood lipidome. We see that the differences between people with schizophrenia and healthy people are just not very similar in the cohorts. We found many statistically significant differences in one data set that were not found in another. Our approach was just to see which differences are common to all the three cohorts. Our hypothesis, in my opinion, is quite reasonable: if there are any common changes, then perhaps they reflect something related to the psychiatric disorder itself, rather than some more culture-related or mediated factors. We can see that if we analyze the common changes, they are very well reproduced in different data sets. In addition to these three cohorts, we recruited a fourth cohort here in Russia. These were new patients, and we saw very good reproducibility there.

O.K: You mean that all the three cohorts demonstrated clear differences between patients with severe psychiatric disorders and healthy people – and you could see them. But were there any between the diagnoses themselves? Did patients with schizophrenia, for example, stand out from all the others? 

A.T: Initially, of course, we were very interested in evaluating the differences between various psychiatric disorders. This is a more pressing problem, especially in psychiatry: it is not so difficult to distinguish a person with a severe psychiatric disorder from a completely healthy person; this task does not require complex molecular tests. However, we found in the course of our study that its design did not allow us to accurately distinguish any differences between diseases. We focused specifically on schizophrenia and the differences between people with schizophrenia and healthy people. They were represented by a large data set, and here we were able to draw some conclusions. Based on what we saw in our data, there were no striking differences between various psychiatric disorders. For example, between bipolar disorder and major depressive disorder. This does not mean that there are no differences. It means that we will need to make a more detailed analysis. Our study focused on a somewhat different problem.

O.K: Do you intend to continue looking for these differences? 

P.Kh: Of course we do. I think we should once again start with the brain. This is surprising, but there have been no comparative studies of changes in the lipid composition of the brain in different psychiatric disorders. This is one of the projects we are currently working on. Naturally, we are going to continue our research on plasma, but here again, as in brain research, we depend on psychiatrists. If they provide us with a large number of well-annotated samples, we will be happy to continue.

O.K: I think you have already convinced the psychiatrists, with your ninety percent probability.

P.Kh: I think many people would be very skeptical about this, and rightly so. This is a balanced system: about half with a diagnosis and half without a diagnosis. When we analyze the general population, where the prevalence is about 1% or less, we have a completely different probability calculation. Nevertheless, lipids appear very promising, in my opinion, and it makes sense to continue such studies. It is quite possible that when we fully understand this problem, or at least better understand what is going on, we will see that many of the conclusions we are making now were wrong from the beginning. The fact remains that we see differences in the lipid composition between people with a diagnosis and without one. Both in the brain and in the plasma. But in order to understand the differences between diseases and what exactly they reflect, we just need more research. Of course, we plan to do that. If anyone reading this text wants to work with us, please let us know. We would be very happy. 

O.K: It would be a finer tuning. I have an even more general question for you, Philipp. You also study the evolution of the brain. You have an interest in evolutionary biology. The evolution of the brain and the development of psychiatric disorders in humans, no matter, which specifically, are they related, in your opinion? 

P.Kh: It’s not just me, our entire lab is involved in this. What is the point here? When we look at all these molecules, whether they are proteins, lipids, RNAs, and measure these thousands and thousands of different types of molecules, these data per se tell us nothing. Because they always have to be correlated with something. We need some kind of database to make comparisons with. If we study a disease, we make a comparison with healthy people. Why do we do this? To find the signs that distinguish a healthy brain from a diseased one, and then try to formulate a hypothesis: what went wrong? What was the mechanism of the disease? Accordingly, it is the same with evolution. We are trying to understand how our brain works. In fact, there is no fundamental difference from the point of view of the study design itself. Either we compare a healthy brain and a brain with a breakdown in order to understand what has happened, or we just compare a brain that can, for example, talk, communicate, think very productively, and the monkey brain, which is not so different from the human brain, but in which these abilities are much less developed. From the monkey’s point of view, we are probably all mad. If they could investigate us, no doubt they would investigate us along with some mad monkeys. We do the same.

O.K: There is also the question of whether this is a breakdown; for example, everyone knows the proverbial thin line between genius and insanity.

P.Kh: There are different ways to put it. The main thing is that changes do occur. Some changes in the manifestation of the ability, right? Therefore, if there are changes in behavior, it is important to know what has happened inside, what changes in molecular composition have occurred. Why are we focusing on lipids? It’s not about lipids actually. In any biological system, as the guys said, everything is connected. It’s not that lipids have changed and everything else has not. Lipids simply reflect one aspect of this whole complex, interconnected system. When we put all the data together, the picture starts to crystallize, so it makes a lot of sense to look at different aspects of complex systems like the human brain, in areas like psychiatry. I think it is important to integrate all this data, because when we get it together, we are able to see a clearer picture. 

O.K: Thanks. My last question is: what would you recommend to Russian research groups? What do you  need to do to get published in such highly rated journals?

E.S: One of the obvious tips is to build on the experience gained by your colleagues working in similar areas. Review the literature, see what has already been done in this area, which aspects have been investigated and which have not, and what new can be introduced. As Anna and Philipp have already mentioned, we tried to combine three different samples in our study, which had never been done before. It is important to see if there is any free niche. I would advise young scientists not to be afraid to go into a new field, not to be afraid to be daring, to try something new, to try themselves in large studies. When you have just graduated from University, it is a bit scary to go into the big science that really has some social meaning. You feel that you have an impact on the society in which you live. It is exciting, and can stop you from developing in this direction.

A.T: I also think that collaborations are a key factor. Without them, we could not do anything. The main aspect of this work is that there have been many collaborations not only in Russia, but also with other countries. This is difficult. It’s hard for me to see how such large organizations can be brought together. If this works out, then there is hope for a good work that is of interest to the international scientific community.

P.Kh: This is always a very big job. The guys spent a lot of time on this. You can’t prepare an article like this within half a year. Such studies take at least several years, because everything needs to be organized, collected, measured, analyzed, and then verified.

O.K: And how long did it take you? 

P.Kh: This project took approximately 5 years. Actually, it is really important to get together people with different expertise. In this case, the project would be impossible without psychiatrists. And not only in terms of sample collection, but also in terms of assistance with the interpretation of the results. It always takes an interaction. In addition, of course, you’ve got to try something new. We could just as well take the same samples and investigate what everyone else is investigating, do something like DNA sequencing, and publish a paper. But the results would not be that new. As Lena said, you’ve got to be daring and venture in new directions. It can be scary. In fact, we could easily have ended up with nothing. I initially estimated the probability that we would not find anything as 90%. 

O.K: It turned out quite the opposite. 

P.Kh: It did turn out to be the opposite, but it was more about luck.

O.K: Well, thank you very much for the conversation. I once again congratulate you on this achievement. I think that the influx of endorphins and dopamine that you have received is a good reward and a source of inspiration for your subsequent research.

P.Kh: Thanks to all who participated in this study. The guys have done a great job. If you look at the list of authors, it consists of 51 people. This was a great collective work that resulted in an interesting and, I hope, useful publication.

O.K: Thanks!

P.Kh: Thank you!