6 August 2012. In a July 11 press release, Eli Lilly and Company announced that its metabotropic glutamate receptor agonist pomaglumetad methionil (also called LY2140023) has failed a Phase 2 clinical trial designed to evaluate its efficacy at improving positive and negative symptoms of patients with schizophrenia. Although the full data from this study—the third Phase 2 trial of monotherapy with this drug—are yet to be released, we do know enough to be disappointed: in contrast to the antipsychotic risperidone, the Lilly drug was no better than placebo at reducing symptoms.
The question of what this means for further research with mGluR2/3 drugs draws mixed responses from the schizophrenia research community. Some researchers wonder if the drug could still have positive effects in a subset of patients, or on a subset of symptoms. "Even if the primary endpoint didn’t meet criteria, you then go back and see if there are other symptoms that seem to have responded," said Daniel Javitt, of the Nathan Kline Institute.
In contrast, Bita Moghaddam of the University of Pittsburgh, whose animal experiments provided the initial rationale for the development of the drugs (Moghaddam and Adams, 1998), does not advocate for following up on direct mGluR2/3 receptor agonists as monotherapy, noting, “Once there are clinical data out there, you cannot argue with that.” She does, however, suggest there may be other ways to effect beneficial changes in glutamate neurotransmission in schizophrenia.
In this article, SRF looks back on the effort to develop a truly novel medication for schizophrenia, and asks some experts their opinions on the future of this enterprise.
The Great Glutamate Hope
Agonists of metabotropic glutamate receptor subtypes 2 and 3 have been on the schizophrenia stage for nearly 15 years, ever since Moghaddam’s 1998 study demonstrated that this class of drugs could reverse the cognitive deficits and psychosis induced by phencyclidine (PCP) exposure in an animal model. However, previous Phase 2 trials of the current drug have been mixed. The first extremely promising study reported a large reduction in symptoms after four weeks’ treatment with LY2140023 (a prodrug that is metabolized into the bioactive compound LY404039), assessed via Positive and Negative Symptom Scale (PANSS) change from baseline, an effect comparable to the antipsychotic olanzapine (see SRF related news story). However, in a second four-week trial, treatment with neither the mGluR2/3 agonist nor olanzapine was more effective than placebo at improving PANSS score. A greater-than-average placebo response was blamed for the lack of effect of both drugs, and thus, the second study was considered a wash (see SRF related news story; Kinon et al., 2011).
The current study, which began in early 2010 and concluded in May 2012, was similar to the design of the previous two, except that the drug was administered for seven weeks. Approximately 1,000 patients with an acute exacerbation of schizophrenia were given placebo, risperidone (2 mg twice daily), or LY2140023 (either 40 or 80 mg twice daily). Unfortunately, neither dose of the mGluR2/3 prodrug met the primary endpoint of the trial—a significant improvement in PANSS score over placebo—although risperidone did. The same was true regardless of whether all schizophrenia subjects were considered, or whether only a genetic subgroup (based on previous studies) was examined. LY2140023 was well tolerated and, importantly, no new safety information was reported.
The first two studies were consistent in terms of response to the Lilly drug. What differed was the response to placebo, providing hope that the previous failed study had been driven by the elevated placebo response. The expectation was that this third trial would replicate the original, bringing us one step closer to a rationally designed treatment for schizophrenia. But the current findings—that risperidone separated from placebo but LY2140023 did not—call into question the efficacy of the mGluR2/3 agonism mechanism, and may have put the problematic placebo hypothesis to rest.
The reaction of the community has been one of great disappointment. “This is obviously a discouraging finding,” said Yale University’s John Krystal. “And we all know that we have a tremendous need for the development of new medications for schizophrenia that target novel brain mechanisms.”
For Jeffrey Lieberman, of Columbia University, New York, the results were very surprising. “In many cases drugs fail … because of mistakes that are made on the part of the pharmaceutical industry or because of some kind of flaw in the development strategy. But this result comes about through no fault of Lilly,” he told SRF. “The theoretical rationale on which this compound was developed is very sound, and there were very strong preclinical data … indicating that it should be effective."
With one positive study, one inconclusive study, and now one failed study, the question is, Which one reflects reality? said Krystal. Of note, in the first study, patients on placebo actually got slightly worse across the trial, making the comparison to the mGluR2/3 drug look more favorable and raising concerns that the positive finding may have been “unrepresentative,” he added. In addition, Krystal said, the current data indicate that there’s a lot we still need to learn about the mechanism of the mGluR2/3 drug, and about the mGlu2 and mGlu3 receptors at which it acts. The two receptor subtypes have different functions, and thus, further basic science studies are needed to better understand the conflicting clinical data.
Hope springs pharmaceutical
In an e-mail, Lilly spokeswoman Keri McGrath told SRF, “We are disappointed by the results. However, Lilly's innovation strategy, which is based on advancing a pipeline of approximately 60 molecules currently in clinical development, does not rest on the success or failure of any single compound.” For now, Lilly says that they are pressing forward with the development of the mGluR2/3 agonist. A Phase 3 trial of the drug is ongoing, with results anticipated following its completion in February of 2013.
Although the current data on mGluR2/3 agonism as a monotherapy are not encouraging, it is possible the drug may work as an adjunct therapy to currently approved antipsychotics. In fact, preclinical data have suggested a synergism between these two classes of drugs (see SRF related news story). Lilly is currently awaiting results from a recently finished, long-term study (16 weeks) examining the efficacy of LY2140023 in addition to atypical antipsychotics in schizophrenia patients with prominent negative symptoms. Another possibility, says Krystal, is that the mGluR2/3 agonist may only be effective in a subgroup of patients, perhaps those who are early in their illness course, consistent with a clinical trial of another glutamate-enhancing drug, sarcosine (see SRF related news story).
Lilly is currently in the process of using a pharmacogenetics approach to identify subgroups of patients who may benefit from LY2140023. Thus far, 23 single nucleotide polymorphisms (SNPs), the majority of them located in the serotonin receptor 2a gene, have been significantly associated with mGluR2/3 agonist response to the PANSS (Liu et al., 2012). Presumably, it is these same SNPs that were used to define the genetic subgroup in the current study. However, the lack of mGluR2/3 agonist efficacy in this subpopulation of patients noted in the press release argues against this approach.
Javitt suggests that Lilly may need to keep searching to find the right indication for this compound. “There’s a lot invested in just demonstrating the safety of the compound, and it sounds like there were no safety issues. So then, often, what you try to do is look through the study to see where a signal is.” For example, one possibility is that LY2140023 may be more effective at improving cognitive deficits as opposed to positive and negative symptoms, consistent with early clinical data in healthy subjects demonstrating that mGluR2/3 agonists can improve working memory deficits induced by ketamine exposure (Krystal et al., 2005). However, it does not appear that the Phase 2 studies have made any cognitive measures, and thus, it will not be possible to address this issue with the current data.
Since efficacy with LY2140023 treatment was achieved in the acute study but not in the most recent one of longer duration, perhaps an allosteric site away from the main ligand binding site would be a better target. Moghaddam was not surprised at all by the current results, saying, "The initial data were from short-term exposure, and this being a direct agonist, it would be expected that this particular ligand … may not work for chronic exposure. Once the actual raw data are made available, if there’s something there early on that disappears, then we may still say that this particular receptor is a good target, but we need to go with allosteric modulation of it.”
Javitt also wonders if mGluR2/3 agonism may be better suited as a treatment for bipolar disorder, since increasing glutamate release through blockade of NMDA receptors produces hyperactivity in rats (Toth and Lajtha, 1986), perhaps for agitation or anxiety rather than psychosis.
The implications of this failed trial reach beyond schizophrenia, says Lieberman, and “reflect a difficulty in developing not just psychotropic drugs for mental disorders … but mechanistically novel drugs that may be trying to become first-in-class and forge a whole new therapeutic strategy.” Clearly, the animal models that we have for screening these compounds for efficacy have limitations, and once the data are released, says Javitt, it will be time to go back to basic science to see if we can use the present results to refine the models.—Allison A. Curley.