NIMH Experimental Therapeutics Paradigm Seeks to Revitalize Intervention Research

More than 50 years ago, biophysicist John Platt published in the journal Science a seminal article titled “Strong Inference,” asserting that the greatest advances in science derive from hypothesis-falsifying experiments—studies designed to exclude, or falsify, one or more hypotheses and then are replicated to winnow out remaining hypotheses.

In this way, a study with a negative result can be enormously clarifying, forcing progressive changes and refinements in theory.

It is a process exactly the opposite of the way that much research on schizophrenia, especially drug discovery research, has been undertaken. What has been lacking in many of these studies is a testable hypothesis about how a molecule might affect a biological target related to schizophrenia.

The National Institute of Mental Health (NIMH) wants to change that. The experimental therapeutics paradigm, initiated under former Director Thomas Insel, M.D., requires that researchers seeking NIMH support for clinical trials in any area, including schizophrenia, demonstrate that the intervention they want to test—whether a drug, a psychosocial intervention, or a somatic therapy—has an effect on some measurable target mechanism believed to be relevant to the disorder being studied.

“We remain deeply committed to intervention research, particularly in areas like schizophrenia where there is tremendous need,” said Joshua Gordon, M.D., Ph.D., who succeeded Insel as NIMH director in 2016. “But we have spent a great deal of money on ‘me too’ studies of interventions that didn’t work, producing negative result after negative result, while never scientifically ruling out or confirming an underlying hypothesis. We were throwing money down the drain.”

It doesn’t have to be that way. As Gordon pointed out in an interview with Psychiatric News, even studies that produce negative results can be powerfully informative if researchers measure some biological target related to the disorder of study.

As an example, stress is known to play a role in psychiatric illness. A researcher might test an antagonist to the corticotrophin-releasing factor (CRF) system (known to be involved in the release of the stress hormone cortisol in the bloodstream), with the hypothesis that doing so will prevent or reduce the severity of depression. If the experimental drug has been shown to blunt the release of cortisol, the researcher can then test the molecule, at a variety of doses, in individuals with depression.

Even if the study is negative, showing no effect on depression, the study will have revealed something vital: blocking CRF is not a mechanism crucial to preventing or reducing depression. “If you show engagement of a biological target with your treatment and prove it doesn’t produce a clinical effect, then you have evidence against the target as a mechanism in the disease,” Gordon said.

The crucial question is, what constitutes a “measurable target”?

“We have been fairly explicit that the measurable target should be as close as possible to the actual posited mechanism of the drug or psychosocial intervention,” Gordon said. “That is, if you are using a drug in a clinical study that binds a particular receptor, the best possible measurement demonstration would be receptor binding. We want the measure to be as proximal as possible to the target.”

However, when it is not possible to assess whether the precise hypothesized target has been engaged, a secondary or downstream effect might be acceptable as a measure, Gordon said. For instance, a researcher who is testing a glutamate receptor agonist but is unable to directly measure receptor binding may, alternatively, show that activation of this receptor (through a downstream pathway) enhances a physiological measure of brain activity, such as the strength of gamma oscillations. (When neurons synchronize their firing, the rhythm of that firing is reflected as brain oscillations; a gamma oscillation is a neural oscillation of a particular frequency that is believed to be involved in cognition.)

“Quantifying gamma oscillations would be a potentially acceptable measure of target engagement in that case,” Gordon said.

Gordon said the institute has never and would never require that treatment or other studies focus only on genes, or, in particular, only on single genes for treatment targets. “We fund plenty of work on pathways that may yield novel treatments. “

What is crucial, he said, is that investigators targeting those pathways be able to demonstrate that they have successfully engaged them so that the study is informative regardless of whether the results are positive or negative.

How can researchers seeking NIMH support for testing psychosocial interventions identify a measurable target?

Gordon pointed out that the majority of clinical trials at the institute are testing psychosocial strategies. “Part of the reason for this is that the psychosocial research community has bought into the idea of target measurement more rapidly than the psychopharmacological community,” Gordon said.

He said a biological, brain-based target mechanism is preferable, but isn’t always possible. What’s crucial is that the researcher identify some measurable target mechanism by which the underlying hypothesis of the study can be tested.

For example, the institute is funding several studies looking at the effect of cognitive-enhancement training for high-risk youth on real-world functional outcomes in schizophrenia. In each case, the measurable targets include adoption of specific cognitive and/or social skills; if the researcher can show that the skills are adopted, then the hypothesis will be reliably tested, whether the actual effect on functional outcomes is positive or negative. The institute is also funding studies of technological interventions for the treatment of schizophrenia (see box at left).

Gordon emphasized that the institute remains committed to intervention research and to drug development for schizophrenia. And he offered some advice—and a challenge—to those seeking NIMH support: “Work with your program officer to figure out what would be an appropriate target measure. If you can’t define one, I want to hear about it. It’s possible we are preventing meaningful clinical trials because there isn’t an appropriate target to measure, but so far no one has taken me up on the challenge.”

He added, “The experimental platform is a way to ensure that the studies we conduct on interventions are rigorous science that will advance the field regardless of whether the experimental intervention shows a clinical benefit.” ■