Depomed CEO Pelzel Discusses Serada's Cold Flush and Placebo Issues

Depomed announced earlier this week that its investigational non-hormonal therapeutic option for the treatment of menopausal hot flashes, Serada (formerly known as DM-5689), failed to meet its primary efficacy endpoints in its two pivotal BREEZE clinical studies. In my opinion, this statement is somewhat misleading: it is not so much that Serada demonstrated limited efficacy in reducing the frequency and severity of hot flushes — but more that the placebo arm demonstrated higher than expected drops in daily frequency of hot flashes.

Serada is an extended release formulation of generic gabapentin, an anti-seizure drug with published clinical data (going back to 2000) showing benefit in helping postmenopausal women control hot flashes. For many women, however, the trade-off has been that the recommended gabapentin dose of 1800 mg (in divided, daily doses) is often associated with unpleasant side effects, including nausea, daytime sedation, and dizziness. The Serada difference — the proprietary Acuform drug delivery technology — is a polymer-based delivery system that allows for targeted, controlled delivery of Serada to the upper gastrointestinal (GI) tract while minimizing the amount of drug that passes through to the lower GI tract.

In a telephone conversation on Wednesday, CEO Carl Pelzel told me Acuform’s unique swelling polymers allow the gabapentin tablet to be retained in the stomach for approximately eight to nine hours, with potential to provide patients with meaningful benefits similar to estrogen (but without the increased risk of cardiovascular disease seen in the Women’s Health Initiative study) and resultant better tolerability and more convenient dosing than generic gabapentin.

Keeping in mind the positive clinical data from Phase 2 Serada data presented at a September2008 NAMS Meeting, I peppered CEO Pelzel with questions on potential causes for the unexpectedly high placebo effect in the two Phase 3 trials, particularly at week 12 in Breeze-1, which reported a dramatic 62 percent drop in frequency of hot flushes from 11.3 to 4.3 per day in the placebo arm:

1. All study participants used electronic diaries to record the occurrence and severity of hot flashes. How compliant were patients in recording actual numbers of hot flush events? Pelzel told me the company is currently collating the data from each PDA.
2. Could the number scale of measurement used to describe the “severity” of a hot flash be too subjective in nature? That is, was it possible some data was corrupted by “acquiescence bias”  (responders feeling the need to rate outcome on anticipated answers they felt researchers wanted to hear) or by the “subject-expectancy effect” (participants are motivated to be in the trial by the belief or expectation the treatment will lead to recovery or relief of symptoms)? To some extent, the study design accommodated for such biases, said Pelzel, but the realized “placebo effect” was still 10 percent to 15 percent higher than expected.
3. I’m no biostatistician, but I remember from my graduate days that the placebo effect must be differentiated from something called “regression to the mean.” In medical parlance, that’s the “natural progression effect,” the statistical likelihood that a sick person would get better absent any treatment. Current data from the Serada Phase 3 trials has only posted efficacy results (from baseline) at week 4 and week 12, yielding little useful information about how participants in the placebo arms fared over time. Pelzel said the company will be collecting as many of the response variable time points as possible. Ergo, a change-score analysis with more time points would help to empower the statistical model: the more that is known about how each person changed over time, the more variances the company could pull out and ascribe to this “placebo effect.”
4. Pelzel also said Depomed is considering re-evaluating the data after using a two-week run-in period — the elapsed time before a trial is commenced when no treatment is given to participants in the study — to weed out potential noncompliant subjects and placebo responders. Such run-in periods aren’t without controversy in research circles, as run-in periods can dilute or enhance the clinical applicability of the results of a clinical trial, depending on the patient group to whom the results will be applied.
5. I also asked Pelzel about the implications of patient selection: Age, years of education, number of years postmenopausal, preexisting cardiovascular disease, sedentary versus active lifestyle, socioeconomic status, follicle-stimulating hormone levels, body mass index, and mental state — research findings suggest all these characteristics can influence severity of hot flushes. The influence of possible co-factors and cohort effects on outcomes data have yet to be completed, Pelzel told me on the conference call.

So what does the future hold in store for Serada? Next steps, said Pelzel, include further analysis of study results (as mentioned), a likely meeting with the FDA in December to discuss either a refinement of the study design (such as one active drug arm instead of two) or additional clinical work in 2010 (which would likely delay an approval review by 12 to 18 months).

In my opinion, Serada’s failure to achieve statistical significance was more an indictment of uncontrolled probability errors dealing with study design than the actual efficacy of the drug. For this reason, I am not comfortable (yet) calling Serada a drug failure. Until either the FDA outright rejects Serada as a hot flash treatment or Depomed walks away, I prefer to declare the Breeze trial results as “inconclusive.”

In my next post, I’ll look at Depomed’s financial ability to survive this hot flash meltdown. Also see my BNET Pharma colleague Trista Morrison’s similar take on the Serada results.