As part of our ongoing coverage of the 2009 International Congress on Schizophrenia Research (ICOSR), 28 March to 1 April 2009, in San Diego, California, we bring you a meeting missive from researcher Tony Altar, Neurodrug Consulting, on a troubling development in schizophrenia trials.
17 April 2009. Increased placebo responses and diminishing efficacy in many antipsychotic drug trials present a serious issue for clinicians and drug developers. A report from Eli Lilly and Co. at the International Congress on Schizophrenia Research in San Diego, and recent publications on this topic, reveal an alarming trend for antipsychotic trials that until recently was a troubling burden mainly for antidepressant trials. The increase in clinical response for the placebo-treated patients in many, though not all, antipsychotic trials occurs in the U.S. as well as abroad. It would appear that sponsor's trials for new compounds in schizophrenia will require a better understanding and resolution of this issue. Clues as to why this unwelcome trend is occurring can probably be found in 1) comparisons of studies that did, and did not, generate such high rates, and 2) similar changes over time in recruiting patients for antidepressant and antipsychotic studies, since both diseases have produced increasing placebo responses over the years. Opinions are also hereby solicited from those Schizophrenia Research Forum readers who can elaborate on these and other possible explanations.
At an ICOSR morning symposium on March 29, Bruce Kinon from Eli Lilly and Co. presented the long-awaited results of a dose-range study of the metabotropic glutamate receptor (mGluR2/3) agonist prodrug, LY2140023 monohydrate, in patients who failed to respond to prior antipsychotic drugs. As we know from the equally anticipated report from Lilly several years ago, LY2140023 had been shown at a twice-a-day dose of 40 mg to treat schizophrenia over a four-week trial, and to an extent about equal to that of 15 mg olanzapine given daily. Both drugs were clearly superior to placebo in that initial trial, where placebo-treated patients got slightly worse (five points on the PANSS Total score) over time.
Such was not the case in the data presented here, a placebo-controlled, double-blind study of 669 patients, termed the HBBI study. LY2140023, at 5, 20, 40, or 80 mg, twice a day, and the daily dose of 15 mg olanzapine, improved the PANSS-total score with increasing effect over the four-week trial, and to about a similar degree. But there was no clear dose-response effect within the 16-fold dose range for LY2140023, and an even greater effect was observed in the placebo group than for any dose of LY2140023 or for olanzapine. A dose-response relationship for LY2140023 would have provided some pharmacological evidence of a cause-effect relationship, but that non-effect, and the strong placebo response, must be reconciled with the body weight increases that were anticipated and did occur with olanzapine, and the finding that 17 of the clinical sites contributed 25 percent of the patients and a large number of the placebo responders. When data were reanalyzed without those sites, the PANSS-positive scores showed at least the same order of effect seen in the initial clinical trial: placebo < LY2140023 < olanzapine, though statistical significance was not obtained.
The increasing placebo response in schizophrenia trials over the last decade has been documented by (Kemp et al., 2008). Figures 1 and 2 of this article show that greater improvements and less worsening in placebo patients are combining with reductions in active responses to 10 antipsychotic drugs over time. These trends have diminished significant changes in antipsychotic drug trials. The paper is co-authored by pharmaceutical and academic organizations, and lists some possible causes.
Remarkably, a 24-point improvement in PANSS Total of the placebo group of a palliperidone trial reported at the ICOSR meeting (Canuso et al.; Ortho-McNeil Janssen Scientific Affairs), and the large placebo improvement in the LY2140023 study reported by Kinon, fall pretty much along or even exceed the Figure 2 downward regression line in the Kemp et al. (2008) paper. In a poster from the 2008 NCEDU meeting (download NCEDU meeting abstracts), Szegedi et al. plotted a very similar graph showing the increasing placebo response for the last 10 antipsychotics tested, and boldly inserted a final column for the LY2140023 placebo but left the point blank, since the trial hadn't yet been reported. Interestingly, the five-point worsening of the placebo group in the first LY2140023 trial sits well above the regression line for this figure, so it appears more will be learned from a careful analysis of the two LY2140023 trials for any procedural differences. Not all antipsychotic studies at ICOSR provided large placebo effects, however; clinical trial reports of PANSS Total values showed little placebo response in results with lurasidone (Poster 81 by M. Ogasa et al. and Poster 83 by J. Guarino et al., both from Dainippon Sumitomo Pharma America) or armodafinil (Poster 187 by J. Kane et al., Zucker Hillside Hospital, New York).—C. Anthony Altar.