Groundbreaking fMRI study finds 4 distinct neurological subtypes of depression

Jessica Hall

New research from Weill Cornell has isolated four distinct neurotypes of depression. But its knock-on effects are much wider in scope. The work establishes biomarkers for depression, and it sheds new light on the physical underpinnings of psychological disease.

The study captured fMRI brain scans from more than a thousand participants, in order to answer a question: What’s different between the brains of healthy people and those with depression? What it found is that within the umbrella category of “people who have major depressive disorder,” there exist (at least) four distinct neurotypes, each with its own cluster of associated symptoms. And the neurotypes aren’t random. They align with their symptom clusters along two major axes: anxiety and anhedonia (anhedonia refers to the inability to feel pleasure). The authors refer to the axes as a shared pathological core, by which we can understand the relationship between brain connectivity and the symptoms of depression. These newly discovered patterns of abnormal connectivity are biomarkers for depression: something neuroscience has been chasing for a long while, without much success.

Neurotypes 1 through 4, in this figure called clusters 1 through 4. This study maps each neurotype in two dimensions: anxiety and anhedonia. At left, a scatter plot of subjects classed into each neurotype. At right, demonstration of how tightly subjects tended to cluster around each neurotype, shaded by number of participants (pale areas denote lower frequency of symptom overlap, while yellow places denote lots of participants with very similar symptoms). Figures: Liston et al, 2016

From the paper (emphasis ours):

We found that, superimposed on this shared pathological core, distinct patterns of abnormal functional connectivity differentiated the four biotypes and were associated with specific clinical-symptom profiles. For example, as compared to controls, reduced connectivity in frontoamygdala networks, which regulate fear-related behavior and reappraisal of negative emotional stimuli, was most severe in biotypes 1 and 4, which were characterized in part by increased anxiety. By contrast, hyperconnectivity in thalamic and frontostriatal networks, which support reward processing, adaptive motor control and action initiation, were especially pronounced in biotypes 3 and 4 and were associated with increased anhedonia and psychomotor retardation. And reduced connectivity in anterior cingulate and orbitofrontal areas supporting motivation and incentive-salience evaluation was most severe in biotypes 1 and 2, which were characterized partly by increased anergia and fatigue.

To parse the results of this study, it’s helpful to know the vocabulary. The frontal cortex, also called the forebrain, is associated with executive control: It’s what lets a kid prevent himself from reaching for a cookie. It sends inhibitory signals and provides a filter between what we think and what we say. The limbic system consists of a set of brain regions related to emotion. I often refer to it as the lizard brain, because the limbic cortex was the earliest cortex to emerge, it probably did so in reptiles, and it handles deep motivations like fear and affection. The amygdala is part of the limbic system, and it particularly handles fear. Also part of the limbic system is the deeply buried anterior cingulate gyrus and the orbitofrontal cortex behind the eyes, both of which handle anticipation and motivation; wacky connectivity here can result in the “don’t wanna,” the feeling that you don’t have enough energy for whatever’s coming next. Similarly, the link between the striatum and the forebrain enables reward processing and the initiation of physical motion. When the forebrain has too much control, people can experience psychomotor retardation: the feeling that gravity has sort of tripled, and everything is just difficult.

Broadly, you can say that some of the anxiety-related aspects of depression are because of abnormal connectivity between the parts of the brain that we feel fear with, and the parts of the brain that exercise control over what we feel. The amygdala says, in effect, “BE SCARED!” Fear of a situation or circumstance can jolt a person into action; this is important when, for example, there’s a hungry lion to be avoided. The forebrain exerts control over the rest of the machine: Immediately dashing away might not be the best course of action, because the hungry lion might see you if you jump up and run, so the forebrain can suppress the signal from the amygdala. This allows feelings of anxiety to be thought through and reasoned past. If the forebrain has too little control over the amygdala during times when there’s no threat of a lion, though, that can result in chronic anxiety. In the same way, if the limbic system isn’t thoroughly connected, or if it’s too inhibited by too much connectivity to the forebrain and striatum, that can result in chronic anhedonia.

Clinical symptom profiles for the four biotypes. Supplementary fig. 2a: Liston et al, 2016

Brain imaging has come a long way from its roots in physiognomy and phrenology, but it’s still frustratingly difficult to line up disorders of the mind with disorders of the brain. The study’s abstract begins, “Biomarkers have transformed modern medicine but remain largely elusive in psychiatry, partly because there is a weak correspondence between diagnostic labels and their neurobiological substrates.” When it comes to things like depression or schizophrenia, for example, there’s no easy single neurological cause, no smoking-gun molecule or lesion that can itself explain each person’s unique symptoms. This is partly because when we defined psychiatric disorders in the DSM, we did that based on symptoms, not necessarily brain anatomy or biomarkers. “Diagnostic labels don’t always line up with the biomarkers,” said lead author Dr. Conor Liston, because “the diagnostic labels weren’t derived from biology in the first place.” The diagnostic mismatch is also partly because despite the fact that you can crystallize these ideas down to too much or too little brain connectivity, brain function isn’t just about the way you’re wired. What we consciously do with the brain also changes how information flows through it over time, by way of neurotransmitters and changing synaptic links.

The study reinforces a portrait of depression as not just a unitary disorder with a gazillion subtypes, but a syndrome: a collection of associated problems and symptoms that can be understood by how they overlap. As it turns out, science agrees with the idea that there’s more than one “legitimate” way to be depressed. Subtypes of depression like bipolar disorder and catatonia still have their own symptom profiles and connectivity problems. But there’s also this shared pathological core to be reckoned with, as another way of understanding the origins and symptoms of depression, and treating the disorder.

Heat maps depicting biotype-specific patterns of abnormal connectivity compared to healthy controls (black), with the relevant areas outlined. Biotype 1: reduced connectivity in frontostriatal/frontoamygdala networks and within limbic system; anxiety, anhedonia, fatigue. Biotype 2: reduced connectivity within limbic system; fatigue, guilt, suicidal feelings. Biotype 3: increased connectivity in frontostriatal and thalamic networks; psychomotor retardation, anhedonia. Biotype 4: reduced connectivity in frontoamygdala networks, hyperconnectivity in frontostriatal networks; anxiety, anhedonia, psychomotor retardation. Fig. 2e: Liston et al, 2016

fMRI technology is both fundamental to this study, and a limiting factor in what it can claim. What fMRIs see depends on how much oxygen there is in the blood in a patient’s brain. Brains use oxygen when they do their work. If different brain regions consistently light up at the same time, that means that they’re both going through oxygen at an elevated rate, so they probably share a functional connection. Areas with too many functional connections, compared to control brains, are shown in warm colors here, and areas that aren’t tightly connected show up in blue because they’ve got a negative correlation in time. Even if two areas don’t have a direct anatomical connection via some fiber tract, they can be connected via some upstream brain region, and show that connection by lighting up at the same time when the upstream region sends out its signal. As a result, fMRI studies are brilliantly suited to tell us about the functional relationships between different parts of the brain. But its resolution in time means that even with the best fMRI machines we have, we still can’t really use fMRI to tell much about the direction of information flow through the brain.

There are, though, some inferences we can make based on the structure and results of the study. The people who participated in this study had tenacious symptoms of depression that have been resistant to treatment and other clinical trials, so these results may not necessarily carry forward into the garden-variety depressed patient’s treatment plan. Medication use didn’t differ between the four clusters that were discovered, which implies that it didn’t change the resting activity of participants’ brains enough to show up across the different fMRI cohorts — but also suggests that patients’ medication has more of an effect on the active, task-based function of the brain than its resting state. And the study found that of the four neurotypes discovered, type 1 had an 82.5% favorable response rate to trans-cranial magnetic stimulation; those patients experienced some relief from their symptoms. This implies that TMS could be a useful adjunct to treatment plans for some patients conforming to the biotype.

The differences between brain activity and brain connectivity leave a lot of important questions that have to be answered before we’ll have a robust model of the human connectome, especially one that we can use to diagnose and treat disease. This is new work, and its methods and assumptions have to be tried in the field by other researchers in order to be accepted as valid. It raises specific questions that weren’t within the scope of the paper. When there’s abnormal connectivity between two regions, is it just because there are too few or too many neurons in the fiber tract? Or are the right number of neurons wired up, but not all of them working right? Could it be because there are pings being lost or echoed somewhere between sender and receiver? Malformed “packets?” Are there similar neurotypes for other psychiatric disorders? How much of this is genetic and how much of it is environmental or derived from life experience? Research will have to integrate our new understanding of the functional and resting connectome with the recently discovered semantic map and our evolving network diagram of the brain, before we’ll be able to answer questions like these with confidence.