In what may be the smallest double-blind, placebo-controlled clinical trials on record, researchers have shown that treating two individuals with drugs aimed at raising brain levels of glycine improved their psychotic symptoms.

They reported their results in the July 3, 2019, print issue of Biological Psychiatry after earlier publication online.

Genome sequencing had shown that the patients, a mother and son, both had a copy number variant (CNV) consisting of a duplication-triplication that contained roughly a dozen genes.

Among those genes, the researchers were particularly intrigued by the gene for glycine decarboxylase, an enzyme that breaks down glycine, which was the only triplicated gene in the CNV region.

Glycine is a precursor for D-serine, and both glycine and D-serine contribute to activation of the NMDA receptor. Insufficient NMDA receptor activation has been implicated in psychotic disorders, and it stands to reason that having extra copies of a glycine-degrading enzyme might lead to such hypofunction.

The genetic results thus fit well with an existing molecular theory of psychosis, Deborah Levy told BioWorld.

Levy is director of the Psychology Research Laboratory at McLean Hospital, and the senior author of the paper describing the trial.

The trial was conducted by randomly alternating between placebo and one of two treatments that were aimed at raising glycine and D-serine levels in the brain.

Both patients continued to take their regular antipsychotic medication, and "in this particular study, there was a clear improvement on top of what you can achieve with ... clozapine in particular," which is considered the gold standard antipsychotic, Levy said. The researchers hope to make D-cycloserine, which was the more tolerable of the two drugs, permanently available to the two patients in the trial.

Other patients could also potentially benefit, though Levy noted that "it's really important to start by identifying what the biology is in the patient... patients with glutamate and NMDA receptor-related mutations might benefit."

The researchers are also trying to make molecular models, including cell culture models with patient-derived cells, and transgenic mouse models with both the entire genomic duplication-triplication, and just triplication of the glycine decarboxylase gene.

Trial tribulations

Beyond the two patients treated in the trial, and others with genetic alterations that could render them sensitive to normalization of glycine levels, the paper illustrates just how much out-of-the-box thinking it takes to bring precision medicine to psychiatric disease.

While animal models, including behavioral models, can bring some useful insights, they are necessarily limited. The hallmark of psychosis, a disordered perception of reality, simply cannot be evaluated directly in rodents.

Meanwhile, clinical psychiatric diagnosis faces several overlapping challenges.

One is that diagnoses are largely based on symptoms, and that different constellations of symptoms can lead to the same diagnosis.

"You can have a room with five schizophrenics who all meet [diagnostic] criteria, but they have nothing in common," first author Alexander Bodkin, director of McLean's Clinical Psychopharmacology Research Program, told BioWorld.

"It is probably quite a number of conditions," he added. "We call them all schizophrenia – but maybe we shouldn't."

The National Institute of Mental Health (NIMH) has become so frustrated with current psychiatric diagnostic criteria, as enshrined in the American Psychiatric Association's (APA) Diagnostic and Statistical Manual of Mental Disorders (DSM), that then-NIMH director Thomas Insel published a scathing critique of the DSM, writing that "unlike our definitions of ischemic heart disease, lymphoma, or AIDS, the DSM diagnoses are based on a consensus about clusters of clinical symptoms, not any objective laboratory measure," after the publication of DSM-5 in 2013. "In the rest of medicine, this would be equivalent to creating diagnostic systems based on the nature of chest pain or the quality of fever."

The NIMH uses an alternative approach, called Research Domain Criteria, to conceptually categorize mental illness – and to evaluate grant applications.

No matter what framework one uses to categorize them, psychiatric genomics are enormously complex. "The brain is more complex than almost anything else in the body," former McLean researcher Uwe Rudolph, who is currently head of the Department of Comparative Biosciences at the Illinois College of Veterinary Medicine, told BioWorld. And the genetics of mental disorders are correspondingly complex, as well as sometimes very idiosyncratic.

The mother-son duo that the team reported on in their paper were an excellent example of that idiosyncrasy: Although they both have the identical genetic alteration underlying their disorders, they had different diagnoses. One had schizoaffective disorder, while the other had bipolar disorder with psychotic features.

GWAS research, Rudolph said, has uncovered two types of genetic contributors to psychiatric disorders. There are rare variants with large effects, and common variants whose effects are exceedingly small.

Trying to utilize that genomic complexity through the calculation of overall risk scores, he added, "has been interesting for science but not biologically informative for medicine so far." Bodkin added that the team's report, too, "isn't the beginning of a tidal wave, at least not immediately."

In the long run, though, even the "inch by inch" motion of psychiatric clinical trials will improve the current situation as described by veteran trialist Bodkin.

"You study new treatments, and you get a handful of patients who do magnificently well who couldn't get better on anything else," he said.

And then, those treatments are shelved.

Patients who respond are a small subset of the total trial population, and psychiatric drug development most often has no way to understand the molecular underpinnings of those treatment successes, and no incentive to conduct the sorts of giant trials that might – or might not – end up having enough statistical power to show an effect, Bodkin said: "It's heartbreaking."