COVID-19 Show message

Skip to content

Post by Former NIMH Director Thomas Insel: Experimental Medicine

By on

In previous posts I have described the crisis of medication development for mental disorders. Medications developed over the past five decades have been prescribed widely but have not been sufficient for reducing the morbidity and mortality of mental disorders. Yet there is diminishing activity in research and development for new medications within either the biotech or pharmaceutical industries. While the development of psychosocial interventions and devices, including the use of mobile technologies, is promising, the absence of a robust development pipeline for more effective medications would be worrisome in any area of medicine and should be a grave concern to the mental health community.

What can we do? Medication development is slow, expensive, and high risk. Indeed, the most recent data suggest that creating small molecule medications across disease areas is, on average, a 15 year1 endeavor that costs over $2B2 and fails more than 95% of the time3 (see figure below). No wonder that industry has reduced their investments. Can NIMH afford to invest its public funds in medication development? Can we afford not to?

The answer to both questions is complicated. To ignore the need would essentially write off “hope” for those not responding to current medications, not only in the short term but over a 15 year time frame. One approach is to improve the pipeline, making it more efficient. The new NIH National Center for Accelerating Translational Sciences (NCATS) will be developing resources, such as rescuing and repurposing existing medications, to reduce the time and cost of medication development. Some of these efforts, such as the recently announced, and rapidly expanding, initiative to make existing compounds available to academic scientists, may be helpful for research on mental disorders.

But the problem for new therapies for mental disorders is not only lack of compounds but lack of understanding of the targets for treatment development. Existing antidepressants and antipsychotics have many proposed molecular targets, but none that have been shown to be necessary or sufficient for their clinical effects. Amazingly, after three decades of broad use of these medications, we still don’t know how they work when they are effective. And we lack effective pharmacologic treatments for too many patients and for some major sources of disability, like the cognitive deficits of schizophrenia or the social deficits of autism. As I have said before, for better treatments, we need better science. For new treatments, we need to identify and confirm targets, as is done already in cancer and other disease areas.

As a result, NIMH is shifting from large clinical trials that promise an incremental improvement to a model called “experimental medicine.” In experimental medicine, drugs are used as clinical probes and the immediate goal is not to develop a treatment but to identify or verify a target. Using proof of concept studies we can determine the ability of the drug to act on a target and affect a biological process or endpoint related to a clinical disorder, such as demonstrating that the new compound occupies relevant neural receptors or produces relevant changes in brain activity. This approach acknowledges that animal studies, while critical for neuroscience, are not consistently predictive of how medications will work in humans, homo veritas. Experimental medicine focuses on human studies rather than rodent research. Clinical studies can be small but they include biomarkers and neurocognitive outcomes to determine mechanisms of action. Even negative results can be informative because if a proposed target is engaged, but there is no effect on a relevant biological process or endpoint, we can rule out that target. For example, antidepressant effects have variously been proposed to involve changes in serotonin neurotransmission, hippocampal cell birth, and changes in stress hormones, among many other effects. By ruling out some targets and focusing on those involved in the biology of the disorder, we can direct treatment development much more efficiently.

Experimental medicine is an experiment. Its success depends on identifying tractable targets. Given that over 95% of compounds fail during the clinical phases of development (a fact not appreciated by looking at the published literature which is biased towards positive results), success may require rapid failures in order to conserve resources by moving quickly to the next candidate (“fast-fail”4). Success will also be determined by how others get involved, from industry to the academic community to patient advocates. We will need public-private partnerships, clinical trial networks, and creation of a new “pre-competitive” culture to change the face of medication development and have more alternatives for people with mental illness.


1 Dimasi JA. Risks in new drug development: approval success rates for investigational drugs. Clin Pharmacol Ther 2001; 69: 297-307.
2 Munos B. Lessons from 60 years of pharmaceutical innovation. Nat Rev Drug Discov 2009; 8: 959-968.
3 DiMasi JA, Feldman L, Seckler A, Wilson A. Trends in risks associated with new drug development: success rates for investigational drugs. Clin Pharmacol Ther; 87: 272-277.
4 Paul SM, Mytelka DS, Dunwiddie CT, Persinger CC, Munos BH, Lindborg SR et al. How to improve R&D productivity: the pharmaceutical industry's grand challenge. Nat Rev Drug Discov; 9: 203-214.