If you want to cure an illness, you try to understand its underlying cause or causes, as a precondition for modifying, reducing, or eliminating them. That, in highly simplified terms, is one of the scientific rationales for much contemporary research on mental illness.

But determining causation is a difficult matter in many illnesses, especially those that are causally complex. Psychiatric illnesses such as schizophrenia, depression, and bipolar disorder are considered highly complex, the product of factors that can span an enormous range, from the unique specificities of an individual’s genetics to those of his or her social (or even pre-birth) environment.

Shan Siddiqi, M.D., a neuropsychiatrist at Harvard Medical School who received a BBRF Young Investigator grant in 2019, was trained as a physician to take care of patients with mental illness. He continues to do so, although he devotes much of his time right now to research, and specifically, figuring out more about causation so that in the future he and others can provide more effective therapies.

In explaining the thrust of his research, and specifically a paper he and colleagues recently published in Nature Human Behaviour—one that may have important implications for improving current therapies—Dr. Siddiqi reflected on a longstanding problem regarding the question of causation in mental illness.

There’s an expression well known among scientists that states: “correlation is not causation.” By this, Dr. Siddiqi explains, researchers mean to say that “there is a big difference between things that are merely correlated”—things that occur together—and situations in which one can prove that this thing is the cause of this other thing.

“We have this problem when we study the brain,” Dr. Siddiqi says. “What we have been doing for a long time is looking at correlates of symptoms. For example, we take a group of people who suffer from major depression and compare them in various ways with a group of people without depression. We hope to see what distinguishes the people who are depressed.

“That’s quite useful sometimes, because some of the factors that correlate with depression—things we don’t see in the non-depressed group—may turn out to be causal factors. At the same time, however, some of the correlates will prove not to be causal.”

The issue is figuring out how to tell causal and non-causal factors apart. It’s a grand-challenge problem addressed in the research Dr. Siddiqi and colleagues have performed with help from his 2019 BBRF grant.

Dr. Siddiqi says the project leading to his newly published paper had its origins in conversations he was having with a senior researcher at Harvard, Dr. Michael Fox. They had access to 14 highly detailed, independently collected sets of brain imaging data from 12 different institutions in the U.S. and other countries. The data documented 713 cases: individuals who had changes in depression in response to various forms of therapy or injury.

“There are so many variables in depression,” Dr. Siddiqi notes. The question was: “How can we compare different people with the same symptoms—knowing that besides the depression symptoms that they share, there are many variables potentially making these individuals hard to compare.” He refers to factors ranging from socioeconomic status to genetics to childhood exposures to other underlying medical or neurological conditions. A similar question about how to properly compare people applies to those who have the same diagnosis—in this case major depression—but report different combinations or severity of symptoms.

At one point, Dr. Fox said to Dr. Siddiqi: “What do you figure would happen if you were to combine every single dataset that we have—all 14? What would you get?”

Dr. Siddiqi’s reply reflected what common sense would suggest: they would likely get nothing that would bear on the problem of causation, since the people covered in the 14 datasets were so fundamentally diverse. Some of the data was about people whose depression was a side effect of deep brain stimulation for Parkinson’s disease; others had received DBS for epilepsy; still other data was about people who had major depression but no other significant co-occurring illnesses; other datasets focused on people whose major depression or Parkinson’s disease was triggered after they had suffered a stroke or an accident that resulted in brain damage.

But both researchers were curious. Dr. Siddiqi ended up saying: “It probably won’t work, but we should try it. If it fails, we can then figure out why and perhaps learn from that.”

A COMMON CAUSAL CIRCUIT

The remarkable thing is that by combining the 14 datasets and comparing them in very sophisticated and rigorous ways, Drs. Siddiqi, Fox and colleagues were able to identify what they believe is a “common causal circuit” involved in major depression, which was consistent across all of the diverse datasets that they scrutinized. Among the co-authors on their paper reporting the results were two members of the BBRF Scientific Council, Mark S. George, M.D., and Helen S. Mayberg, M.D., who are pioneers, respectively, of TMS (transcranial magnetic stimulation) and DBS (deepbrain stimulation), two therapies used to treat depression and other illnesses. Five other recipients of BBRF grants were members of the research team.

“When I looked at the results,” Dr. Siddiqi relates, “I said at first: ‘This can’t be right. Let me see if I can find a way to break it.’”—by which he meant, go back over the work to see if he could find a place or places where the team made an error or missed something. “I spent hours and hours—so many that I have to admit it got to the point of annoying my family. I literally stayed up all night, over days, trying to find ways to see where we went wrong.”

But he couldn’t. And this led to another remarkable moment. Dr. Siddiqi realized that “if we really did succeed in finding a common causal circuit in depression”—this will have to be replicated in follow-up work in his lab and others, which is now under way— ”then we succeeded in spite of the heterogeneity, in spite of all the factors we know that make the people in the 14 datasets unalike.” This, he says, makes the finding especially powerful.

The team studied major depression severity and underlying brain circuit and network dynamics in the 14 datasets, mapping results against “connectome” information—consensus wiring diagrams showing how different parts of the brain are connected, in a typical person.

Five of the datasets comprised a total of 461 individuals who were assessed for major depression after suffering brain injuries; four datasets comprised 151 individuals with major depression who were treated with TMS, a form of non-invasive brain stimulation; and five datasets comprised 101 individuals diagnosed either with major depression, epilepsy, or Parkinson’s disease, all of whom received DBS, which involves surgically implanting a kind of pacemaker in the brain to deliver therapeutic electrical stimulation.

One thing that is interesting about the team’s finding of a “common circuit” in depression is that it draws attention to circuitry that spans different regions. To use the example of the datasets that focused on people who had suffered brain injuries which led to their depression: these “lesions,” as they are called by doctors, occurred in widely separated parts of the brain. There was no single spot in the brain at which, if tissue damage occurred, the patient would then develop major depression. The actual message was: at the many lesion sites analyzed, injury at particular spots was found to impair the function of some portion of a common circuit or network spanning different brain regions.

A second important finding based on the 14 datasets has to do with people with major depression who had been successfully treated with either non-invasive TMS or invasive DBS brain stimulation therapies. Even though these treatments use different technologies, and even though they are not targeted in the same place in the brain in each patient, they were found to modify the function of a common circuit, the study indicated. Remarkably, this circuit was found to be “similar” to the common circuit found in the study subjects with brain lesions that caused depression. It was also “similar” to a circuit directly linked with causality in depressed patients with epilepsy and Parkinson’s disease.

By “similar,” the team means to suggest something stronger than a vague correlation. Their use of the term reflects an outcome, in Dr. Siddiqi’s words, which is much stronger than would be expected to happen by chance. It’s the result of complex mathematical analysis based on where the circuitry being measured in any given patient is arrayed within the three-dimensional space of the brain. A great deal of machine- and computer-guided measurement went into the determination that the “spatial correlation” of a causal circuit identified in people with brain lesions was similar in “robust” ways with the circuit impacted by brain stimulation therapies that alleviated depression in TMS and DBS patients.

For a century or more, physicians treating people who have suffered serious brain damage have documented a range of associated symptoms involving speech, vision, movement, and memory. In some instances, such lesions give rise to psychiatric symptoms, including those of major depression.

In their paper Drs. Siddiqi, Fox and colleagues do note that “our convergent [depression] circuit includes regions previously implicated in depression.” These include the subgenual cingulate, ventromedial prefrontal cortex, and dorsolateral prefrontal cortex. But that is not the primary value of their findings, Dr. Siddiqi suggests.

WORKING BACKWARD FROM LESIONS TO TARGETS

In his view, the most important aspect of the team’s work is that it may be feasible to work backward from brain lesions to find circuits and targets within them whose modification—for example, by TMS or DBS—might help to alleviate a range of psychiatric illnesses. “Our work provides concrete evidence that brain lesions map to treatment targets,” Dr. Siddiqi has noted.

Among the datasets the team studied were those involving individuals with brain lesions that were causally linked not with depression but with the onset of motor symptoms of Parkinson’s disease. A common circuit was found in these individuals. As with the common circuit identified in depression, this circuit in the Parkinson’s patients with lesions was “similar” to the circuit that DBS brain stimulation modified to achieve reductions in Parkinson’s symptoms.

IMPLICATIONS BEYOND DEPRESSION

“This is the strongest evidence to date showing that lesions causing symptoms can identify therapeutic targets for symptom relief,” the team wrote. For this reason, they hope their findings will have “therapeutic implications well beyond depression and Parkinson’s disease.”

Using a circuit mapping method that they used to link lesions in major depression and Parkinson’s disease with circuits that were modified in successful treatments for each illness, the team is now working on circuit maps for mania, hallucinations, movement disorders, as well as addiction, PTSD and OCD.

“On another track, we’re also looking at targeting specific symptom clusters within a disorder,” Dr. Siddiqi says. “In 2020 we published a preliminary paper on this question with regard to treating two clinically distinct manifestations of depression (called dysphoric vs. anxious/somatic depression).”

In their recent paper, Dr. Siddiqi and colleagues make an important point about their research approach which bears on the prospect of improving therapeutic targeting for people with different subtypes of an illness—in this case, depression.

“Our analysis may seem circular given that TMS and DBS targets for major depression were chosen because they were already known [due to therapeutic results] to be part of a ‘depression circuit.’” However, they go on to explain, “the left prefrontal cortex appears as part of our depression circuit not because it has been targeted with TMS but because different TMS targets across the left prefrontal cortex produced different effects on depression.”

Similarly, they noted, “different DBS sites produced different effects on depression symptoms, depending on their connectivity to the left prefrontal cortex. Also, different lesion locations were associated with different amounts of depression, depending on their connectivity to the same part of the cortex.”

Among other things, this suggests that modifying the common circuit in depression (or in other illnesses if they are identified and verified) may end up having different degrees of impact upon patient symptoms, or may impact different symptoms, depending on where and how the circuit is modified.

These are among the many intriguing questions that Drs. Siddiqi, Fox and colleagues intend to address in their ongoing research—and they must perform this work, they say, before they will be ready to suggest specific modifications in current treatment targets for all or subsets of patients in depression or other disorders.

Written By Peter Tarr, Ph.D.

Click here to read the Brain & Behavior Magazine's May 2022 issue