Bitopertin: the good news and bad news.

In this issue, Umbricht and colleagues1 reported improvement of negative symptoms of schizophrenia associated with 2 of 3 doses of the glycine transporter type 1 inhibitor bitopertin (RG1678) in a phase 2 placebo-controlled, 8-week add-on trial. Although the therapeutic effect was only modest, this is very welcome news because the path to drug development in schizophrenia has been littered with disappointments. The example of clozapine, which was synthesized in 1958 and received Food and Drug Administration approval in 1989, provided impetus to develop similarly effective new compounds. Since then, a series of preclinical and clinical studies by Paul Janssen, MD, pioneered the addition of 5-hydroxytryptamine type 2A antagonism to D2 antagonism, which launched risperidone, the first of the newer-generation antipsychotics, and refinement of dopamine D2 receptor partial agonism led to aripiprazole. Both developments achieved a reduction in neurologic adverse effects, but the promise of clozapine has yet to be realized. Other approaches that followed from discoveries in neuroscience have failed or have yet to reach clinical validation. Bitopertin represents the culmination of more than 2 decades of basic and clinical research on the glutamatergic model of schizophrenia.2 The glutamatergic model of schizophrenia followed from the observation that blockade of N-methyl-D-aspartate (NMDA) receptor–gated channels by phencyclidine or ketamine produces symptoms characteristic of schizophrenia.3 Subsequently, convergent support came from animal models, ketamine provocation studies in humans, genetic studies, and postmortem studies. The drugs that were available to test the hypothesis were limited by poor brain penetrance (glycine, Dserine, and D-alanine) or by partial activity (D-cycloserine and sarcosine) and as a group produced mixed results with a tendency for positive effects in early, smaller trials and negative results in later, larger trials.4-6 This lack of consistency may reflect the inherent limitations of these early agents, inadequate dosing, or methodological problems associated with large-scale replication trials such as high placebo response rates, participant heterogeneity, or lack of precision in measures of negative symptoms. The unreliability of clinical trials has also plagued the development of the other leading glutamatergic candidate, LY2140023 monohydrate, an mGlu2/3 agonist that demonstrated significant efficacy in a first trial but was not effective in a subsequent trial in which placebo response was much greater. In addition, unlike first-generation antipsychotics, second-generation antipsychotics may enhance NMDA receptor signaling—positive results with the glycine site agonists came mostly from early studies in which they were combined with first-generation antipsychotics. In short, results have been promising but inconsistent with agents acting at the glycine modulatory site of the NMDA receptor, but a more rigorous test of this approach awaited the arrival of a more suitable compound such as bitopertin. Bitopertin is a highly selective and potent glycine transporter type 1 reuptake inhibitor that increases synaptic glycine levels and thereby facilitates NMDA-channel opening. Excessive calcium influx through the NMDA-gated channel can be neuroxotic—a risk that is counterbalanced by several safe guards that may complicate this therapeutic approach. For example, activation of the glycine site results in downregulation of the NMDA receptor by endocytosis.7 In addition, an endogenous antagonist at the glycine site, kynurenic acid, competes with glycine and D-serine in the modulation of glutamatergic signaling. The team of investigators at F. Hoffmann–La Roche Ltd performed positron-emission tomography and cerebrospinal fluid studies with bitopertin in healthy participants, informed by animal behavioral studies, to determine receptor occupancy and resulting glycine concentrations in the central nervous system; this allowed them to estimate the range of therapeutic concentrations.8 Their elegant preparatory work proved to be critical because they identified an inverted U therapeutic window dose-response relationship and managed to capture this segment of the dose-response curve in the range of doses administered in this study. Their results suggest that roughly 50% occupancy and a corresponding moderate, 2-fold elevation of glycine levels may be optimal. The investigators also carefully selected participants to enrich their sample with primary negative symptoms and excluded patients taking clozapine because of evidence that it may have actions at the glycine site. The results must be considered exploratory because this was a dose-finding proof-of-concept trial that examined several bitopertin doses without statistical correction. The 2 lower bitopertin doses, 10 mg daily and 30 mg daily, produced a moderate improvement in negative symptoms, which was statistically significant in adherent patients who completed the trial but was not significant in the more conservative intent-to-treat analysis, which included dropouts. It is noteworthy that the global rating of negative symptom improvement was also significantly improved, suggesting a clinically relevant effect. On the other hand, bitopertin did not affect measures of cognition or psychosis, which is puzzling because impairments in Related article page 637 Opinion